International Conference on Sustainable Solutions

in
Engineering and Technology
(SSET-2024)

Convener: Dr. Veena S. Soraganvi

Organizing Committee:

Chief Patron : Dr. Veeranna C. Charantimath, Chairman, B. V. V. Sangha,

Bagalkote
Patrons : Shri. Mahesh Athani, Hon. Secretary, B. V. V. Sangha,
Bagalkote
Dr. R. N. Herkal, Director of Technical Education, B. V. V.
Sangha, Bagalkote
Shri. B. S. Haravi, Development officer, BEC, Bagalkote
Convener : Dr. Veena S. Soraganvi, Principal
Organizing Chair : Dr. Mahabaleshwar S. K., Dean (R&D) and ICT
Organizing : Dr. P. N. Kulkarni, Dean (Academics)
Committee Dr. Anil D. Devangavi, Dean (Quality Assurance) & HoD
Department of AI & ML
Dr. S. G. Kambalimath, Dean (Career Guidance and
Placements)
Dr. Bharati S. Meti, Dean (Student Welfare) & HoD of
Biotechnology
Dr. K. Chandrasekhar, Controller of Examinations
Dr. B. R. Hiremath, HoD, Department of Civil Engineering
Dr. Vinay V. Kuppast, HoD, Department of Mechanical
Engineering
Dr. R. L. Naik, HoD, Department of Electrical and Electronics
Engineering
Dr. V. B. Pagi, HoD, Department of Computer Science and
Engineering
Dr. Jayashree D. Mallapur, HoD, Department of Electronics
and Communication Engineering
Dr. Krishnamurthy Bhat, HoD, Department of Electronics
and Instrumentation Engineering
Dr. Sadhana P. Bangarshetti, HoD, Department of
Information Science and Engineering
Dr. C. M. Javalagi, HoD, Department of Industrial Production
Engineering
Dr. V.G. Akkimaradi, HoD, Department of Automobile
Engineering
Dr. Chayalakshmi C. L., HoD, Department of Electronics &
Computer Engineering
Dr. R. B. Tapashetti, HoD, Department of Management
Studies
Smt. Sudha K. S., HoD, Master of Computer Applications
Dr. S. U. Durgadsimi, HoD, Department of Physics
Dr. S. K. Patil, HoD, Department of Physics
Dr. Vidya M. Shettar, HoD, Department of Physics
Dr. B. G. Hokarani, HoD, Department of Physics
Track 2: Electrical, Communication and Networking
External Reviewers
Sl. Name and Affiliation Sl. Name and Affiliation
No. No.
01 Dr. B. N. Patil 01 Dr. Chinmayananda A.
A. G. M. Rural College of Engg. & IIIT, Dharawar
Tech Hubballi
02 Dr. Shivappa Sabarad 02 Dr. B. G. Sheeparamatti
O & M Division, GESCOM Sahyadri College of Engineering
Munirabad Managalore
03 Dr. Mukta Bannur 03 Dr. Ramesh B. Koti
BLDEA’s College of Engg. & Tech Gogte Institute of Technology
Vijayapura Belagavi
04 Dr. Nagaraj B. G. 04 Dr. R. M. Math
Vidyavardhak College of Engineering B. L. D. College of Engg. & Tech.
Mysuru Vijayapur
05 Dr. S. P. Padaganur 05 Dr. Vivek Jaladi
B. L. D. College of Engg. & Tech. Lingarajappa Engineeing College
Vijayapur Bidar
Internal Reviewers
01 Dr. R. L. Naik 02 Dr. Shridhar K.
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
03 Dr. Chayalakshmi C. L. 04 Dr. P.N. Kulkarni
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
05 Dr. B. F. Ronad 06 Dr. Rajani S. Pujar
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
07 Dr. M. A. Sutagundar 08 Dr. Somu P. Parande
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
09 Dr. S. Y. Goudappanavar 10 Dr. A. H. Unnibhavi
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
11 Dr. M. J. Sataraddi 12 Dr. Ajaykumar C.Katageri
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
13 Dr. J. D. Mallapur 14 Dr. Kirankumar B. Balavalad
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
15 Dr. Sarojini B. K. 16 Dr. Kirankumar Y. Bendigeri
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
17 Dr. A.V. Sutagundar 18 Dr. S. B. Kumbalavati
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
19 Dr. K. Bhat 20 Dr. P. M. Channal
Basaveshwar Engineering College, Basaveshwar Engineering College,
Bagalkote Bagalkote
Chairman’s Message

It is my distinct honor and privilege to welcome you to the International Conference

on Sustainable Solutions in Engineering and Technology, hosted by Basaveshwar
Engineering College. This prestigious event brings together thought leaders,
researchers, and practitioners from around the globe to explore innovative solutions
that address the critical challenges of sustainability in engineering and technology.

Our Sangha is deeply committed to fostering an environment of academic

excellence and innovation. We believe that the interdisciplinary exchange of
knowledge and ideas at this conference will lead to meaningful collaborations and
impactful solutions. The topics covered here, ranging from renewable energy to
sustainable infrastructure, are crucial for building a resilient and sustainable world.

The themes and discussions of this conference are more pertinent than ever as we
face the dual imperatives of advancing technology and preserving our environment.
I am confident that the insights and solutions that emerge from this conference will
make significant contributions to our shared mission of sustainability.

I extend my heartfelt thanks to all the keynote speakers and participants for making
this event successful. I congratulate the organizers for their hard work and
dedication.

Dr. Veeranna C. Charantimath

Chairman,
B. V. V. Sangha, Bagalkote
Secretary’s Message

As we navigate the complexities of modern development, it is imperative that we

integrate sustainable practices into every facet of our technological advancements.
This conference provides a vital platform for researchers, practitioners, and
innovators to share their insights, discoveries, and strategies for creating sustainable
solutions that will shape our future.

At Basaveshwar Engineering College, we are deeply committed to fostering an

environment of academic excellence and innovation. Our goal is to drive forward
the boundaries of knowledge and practice in ways that are sustainable and beneficial
for society as a whole. This conference is a testament to our dedication to these
principles and our belief in the power of collaborative effort.

Shri. Mahesh Athani

Hon. Secretary
B. V. V. Sangha, Bagalkote
Technical Director’s Message

I am particularly excited about the innovative solutions and cutting-edge research

that will be presented during this conference. The intersection of engineering and
sustainability presents unique challenges and opportunities, and it is through
gatherings like this that we can share knowledge, inspire innovation, and collaborate
on projects that will have a profound impact on our world.

Our commitment to sustainability is not just a goal but a guiding principle that
influences all aspects of our work. This conference is an ideal platform to explore
new ideas, methodologies, and technologies that can lead to sustainable growth and
development. The diverse array of topics and the expertise of our participants
promise a rich and enlightening experience for all.

I extend my heartfelt thanks to all the participants, keynote speakers, and organizing
committee members for their dedication and hard work in making this conference a
reality. Your contributions are invaluable to the success of this event and to the
advancement of sustainable engineering and technology.

Dr. R. N. Herkal
Director of Technical Institutes
B. V. V. Sangha, Bagalkote
Principal’s Message

Dear Collegues & Researchers

I feel happy to organize International Conference titled “Sustainable Solutions in

Engineering and Technology” in Basaveshwar Engineering College, Bagalkote.
This prestigious event, promises to be a landmark occasion, bringing together
leading experts, researchers, and innovators from around the globe.

Focus of this conference will be on exploring cutting-edge approaches and

technologies that address the pressing challenges of sustainability in engineering
and technology. With a diverse array of topics ranging from renewable energy
solutions and sustainable materials to smart infrastructure and green manufacturing,
we aim to foster collaboration and inspire breakthrough ideas. Researchers from all
disciplines gather here to explore the multidisciplinary approaches in designing and
implementing systems that meet present needs without compromising the needs of
future generations.

Our institution is honored to be the venue for this significant event and is committed
to providing an enriching experience for all participants. We are confident that the
key notes from experts, presentations from researchers will lead to valuable insights
and partnerships that will drive forward the agenda of sustainable development.

I extend my deepest gratitude to Management for their continued support. I thank

all the Keynote speakers, participants and orgnaising committee members for their
continued support and engagement in this crucial event.

Dr. Veena Soraganvi

Principal
BEC, Bagalkote
Dean (R & D)’s Message

Dear Esteemed Colleagues and Participants,

It is with great pleasure and pride that I welcome you to the International Conference
on Sustainable Solutions in Engineering and Technology. This conference is a
testament to our collective commitment to advancing research and innovation in
ways that are both technologically forward-thinking and environmentally
sustainable.

Our institution has long been at the forefront of fostering research that addresses
global challenges. This conference serves as a crucial platform for sharing
knowledge, exchanging ideas, and forging collaborations that can lead to sustainable
advancements. The contributions from our distinguished speakers and participants
are essential for driving forward the agenda of sustainable development in
engineering and technology.

I extend my deepest appreciation to all the researchers, practitioners, and organizers

who have worked tirelessly to make this conference a success. Your dedication and
expertise is the cornerstone of this event, and your contributions are instrumental in
shaping a sustainable future.

Let us seize this opportunity to collaborate, innovate, and inspire one another as we
work towards sustainable solutions that will benefit not only our generation but
those to come.

Dr. Mahabaleshwar S. K.
Dean (R & D) and ICT
BEC, Bagalkote
 CONTENT
Track 2: Electrical, Communication and Networking ...................................260

27. Anti-Sinking Airbag and Indicating System for Vehicle during Flood -
Krishnamurthy Bhat, Chayalakshmi C. L........................................................... 261

28. Combined Effect of Noise Reduction and Multiband Frequency

Compression for Improving Speech Perception in Monaural Hearing Aids on
Source Localization - Jyoti M. Katagi, Pandurangarao N. Kulkarni ................267

29. Data Management in Edge Computing: Opportunities and Challenges -

Piyusha S. Shetgar, Chayalakshmi C. L., Mahabaleshwar S. K., Rajani S. Pujar..
.......................................................................................................................... 279

30. Development and Performance Testing of Automatic Seed Sowing Agri

Robot - Hamjadali A. Umachagi, Prashant Kadi, Pavankumar Kulkarni,
Vinayak Sharma, Saniya M. Patel, Dr. Basanagouda F. Ronad ......................... 288

31. Edge Computing Based Smart Health Care System - Sadashiv Badiger,
Chayalakshmi C. L., Mahabaleshwar S. K., Rajani S. Pujar .............................. 299

32. Integrating Block chain in EV Charging Systems for Secure and Efficient
Infrastructure - Preetam Kanal, Akshata Dhagade, Bhagyashree Belagali, Darini
Budihal, Rajani S. Pujar.................................................................................... 311

33. IoT Enabled Infant Incubator for Healthcare Centers - Sharanappa P. H.,
Basavaraj M. Angadi, Mahabaleshwar S. Kakkasageri, Sudha K. S................... 318

34. Machine Learning Approaches for Data Storage in IoT: A Review - Supriya
B. Harlapur, Mahabaleshwar S. Kakkasageri .................................................... 326

35. Performance Analys is of Inter-satellite Optical Wireless Communication -

Onkar A., Anudeep Daggupati, Shreevatsa Kulkarni, Vineeth Pelliyembil .........337

36. Review of Structural Health Monitoring Methods: A Machine Learning

Approach - Rashmi M. Kittali, Ashok V. Sutagundar ........................................345

37. RFID based Smart Shopping Trolley - Mamata J. Sataraddi,

Mahabaleshwar S. Kakkasageri, Vedant Vanaki, Omkar Mutnal, Satish Bailwad,
Neelakant Vastrad ............................................................................................. 358
38. Smart Home Automation using IoT - Anjali Honakeri, Niveditha M.,
Sweta M. Elangadi, Priyadarshini Jalikatti, Chayalakshmi C. L. ...................... 368

39. Speech Intelligibility Enhancement based on Spectral Splitting Technique -

Aparna Chilakawad, Pandurangarao N. Kulkarni ............................................ 376

40. Survey on Health Monitoring System - Kartik Kulkarni, Shrinidhi Joshi,

Chandrashekhar G. Hadalagi, Shivananda, Anand H. Unnibhavi ..................... 384

41. Towards 6G: An Overview of Next Generation Communication -

Mallikarjun Dheshmuk, Achyut Yaragal, S. B. Kumbalavati,
Kirankumar Y. Bendigeri, Jayashree D. Mallapur ............................................ 390

42. A Comprehensive Review of Resource Management Techniques in Edge

Computing - Shrishial N. Chamatagoudar, Rajani S. Pujar, Mahabaleshwar S.
K., Chayalakshmi C. L. ..................................................................................... 398

43. Applications of Machine Learning and Deep Learning in Farming: A

Review - Halaswamy B. M., Vinutha C. B. ....................................................... 406

44. Aggregation Based on Clustered Data (ABCD) with Edge Computing -

Achyut Yaragal, M. N. Deshmukh, Kirankumar Bendigeri, S. B. Kumbalavati,
J. D. Mallapur .................................................................................................. 416

45. Challenges and Future Prospects of Thin Film Deposition Techniques: A

Critical Review - Vinay Shettar, Sneha B. Kotin, B. G. Sheeparamatti,
Manjula Sutagundar ......................................................................................... 426

46. Designing the Arduino Based Nutrition Feeding Hydroponic System -

Niveditha N. Shahapur, Sapana S. Lalaki, Allamprabhu V. Kolaki.................... 439

47. Multilingual Regional Speech Classification Using Recurrent Neural

Networks - Sunilkumar M. Hattaraki, Ranjeeta Kumbar, Sushma M. Aloor,
Ranjeeta Patil, Ramya Vajjaramatti, Shankarayya G. Kambalimath ................. 447

48. Contactless Care Systems: A Review - Basavaraj Soratur,

Mahabaleshwar S. Kakkasageri, Subhas Meti .................................................. 457

49. Design and Modelling of Droop Control for Small Wind Turbine
Generator in µGrid - Sangamesh Y. Goudappanavar,
Rubinabegaum H. Yadahalli, Megha Sunkad, Prema, Ravindranath ................. 462

50. LoRa based Smart Energy Meter for Theft Detection - Akash P. S.,
Deepa S. Patil, Sagar N. Odunavar, Nagaprasad H., Chayalakshmi C. L. ........ 475
51. UV-C Based Plate Sterilizer - Suhas. R. Hatti, Abhishek C. Arahunasi,
Vijaylakshmi S. J. .............................................................................................. 485

52. Development of PD-PWM Technique for 5-Level Cascaded H-Bridge

Inverter using FPGA for SPV Systems - Basavaraju S. Hadapad,
Raghuram L. Naik, Amrutha H. D., Asha Chougala, Kailash, Madhu D. ...........493

53. A Review of Various Attacks and Detection Methods in Internet of Medical

Things (IoMT) Systems - Benazir Muntasher, Mahabaleshwar S. Kakkasageri
.......................................................................................................................... 504

54. Hybrid SegAN Fuzzy–Unet and DKN Classification for Crop Field
Change Detection using Satellite Images - Ms. Mulik Madhuri Balasaheb,
Dr. Kulkarni P. N., Dr. V. Jayshree ...................................................................514

55. IOT Based Pollution Monitoring and Controlling System - Sarojini B. K.,
Suhas Hatti........................................................................................................522

56. Design and Implementation of Multi Prime Mover Coupled Novel Water
Pump for Small Scale Irrigation Requirements - Vijaykumar Purutageri,
Tejashwini Bagewadi, Basanagouda Ronad, Mukta Bannur .............................. 531
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

Track 2:
Electrical, Communication and
Networking

260
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

27. Anti-Sinking Airbag and Indicating System

for Vehicle during Flood
Krishnamurthy Bhat
Professor,
Department of Electronics and Communication Engineering,
Basaveshwar Engineering College, Bagalkot.
Chayalakshmi C. L.
Associate Professor,
Department of Electrical and Electronics Engineering,
Basaveshwar Engineering College, Bagalkot.

Abstract:

Floods pose significant risks to vehicles, often leading to dangerous situations for occupants
and complicating rescue operations. This paper introduces an innovative anti-sinking
airbag and indicating system designed to enhance vehicle safety during floods. The
proposed system automatically deploys airbags to buoy the vehicle when floodwater
reaches critical levels, preventing it from submerging. Simultaneously, an integrated
indicating system alerts occupants and emergency services about the vehicle's status and
location, facilitating prompt rescue efforts. The anti-sinking airbag system is equipped with
advanced sensors and deployment algorithms to ensure rapid and effective response during
flooding. The indicating system utilizes communication protocols to provide real-time
updates on the vehicle's condition, thereby improving the coordination of emergency
services. This dual-function system not only enhances occupant safety but also contributes
to more efficient and effective emergency responses during flood events. This paper details
the design, functionality, and integration of the anti-sinking airbag and indicating system,
highlighting its potential impact on vehicular safety standards. By addressing a critical gap
in existing vehicle safety technologies, this research aims to pave the way for the
development of more resilient transportation solutions capable of withstanding the
challenges posed by severe flooding.

Keywords:

Anti-sinking airbag, Flood safety, Vehicle safety systems, Emergency response, Buoyancy,
Sensor integration, Real-time communication, Automotive safety technology, Disaster
management.

261
International Journal of Research and Analysis in Science and Engineering

I Introduction:

Floods are among the most devastating natural disasters, causing extensive damage to
infrastructure and posing significant risks to human life. Vehicles, which are essential for
daily transportation, are particularly vulnerable during floods, often becoming immobilized
and leading to dangerous situations for occupants. Traditional vehicle safety systems
primarily focus on accident prevention and occupant protection during collisions, but they
fall short in addressing scenarios where vehicles encounter sudden flooding [1-3].

In recent years, there has been a growing interest in developing innovative solutions to
enhance vehicle safety during floods. This paper presents a novel approach: an anti-sinking
airbag and indicating system designed specifically to mitigate the risks associated with
vehicular flooding. The proposed system aims to provide an automatic response mechanism
to prevent vehicles from sinking when exposed to floodwaters, thereby increasing the
chances of occupant survival and vehicle recovery [4-7].

The anti-sinking airbag system is engineered to deploy when the vehicle detects a significant
rise in water levels, buoying the vehicle and preventing it from submerging. Concurrently,
the indicating system alerts occupants and emergency services about the vehicle's status and
location, facilitating timely rescue operations. This dual approach not only enhances the
immediate safety of the vehicle's occupants but also contributes to a more efficient and
coordinated emergency response.

This paper will delve into the design and functionality of the anti-sinking airbag and
indicating system, exploring the underlying mechanisms and technologies involved. It will
also discuss the integration of sensors, deployment algorithms, and communication
protocols essential for the system's effectiveness. Furthermore, the paper will analyze the
potential impact of this technology on vehicular safety standards and its implications for
future automotive designs. By addressing a critical gap in current vehicle safety systems,
this research aims to contribute to the development of more resilient transportation solutions
capable of withstanding the challenges posed by increasingly frequent and severe flooding
events.

II Literature Review:

The increasing frequency and severity of floods due to climate change have necessitated the
development of innovative safety measures for vehicles. Traditional vehicle safety systems,
which primarily focus on collision prevention and occupant protection, are inadequate in
addressing the unique challenges posed by flood scenarios. This literature review explores
existing research and technologies related to vehicular flood safety, highlighting gaps that
the proposed anti-sinking airbag and indicating system aims to address.

Studies have shown that vehicles are particularly vulnerable during floods, often leading to
life-threatening situations for occupants. According to research carried out by author, many
flood-related fatalities occur in vehicles, with individuals attempting to drive through
floodwaters. Traditional vehicle designs lack the mechanisms to prevent sinking or provide
adequate alerts to occupants in such conditions [8].

262
 Anti-Sinking Airbag and Indicating System for Vehicle during Flood

Current flood mitigation technologies for vehicles are limited. Some advancement has been
made in the area of water-resistant vehicle designs and elevated vehicle structures.
However, these solutions primarily focus on minimizing water ingress rather than
preventing vehicles from submerging. For instance, the development of water-resistant
electronic components discussed in this paper has improved vehicle performance in wet
conditions but does not address the issue of vehicle buoyancy during severe floods [9].

Airbags have been a crucial component of vehicle safety systems for decades, primarily
used for occupant protection during collisions. Research by authors indicated the potential
for adapting airbag technology for other safety applications. However, there is limited
research on the use of airbags to enhance vehicle buoyancy. The concept of deploying
airbags to prevent vehicles from sinking is relatively novel and has not been extensively
explored in existing literature [10].

Effective communication is vital for timely emergency response during flood events.
Advances in vehicle communication systems, such as those discussed by Campolo, have
enhanced real-time data transmission and vehicle-to-infrastructure communication. These
technologies provide a foundation for developing indicating systems that can alert
occupants and emergency services about a vehicle's status during floods. However, there is
a need for integrated systems that combine buoyancy aids with real-time communication
capabilities [11].

The integration of sensors and deployment algorithms is critical for the effectiveness of any
advanced vehicle safety system. Studies by Wang, highlights the importance of sensor
accuracy and rapid response times in safety-critical applications. While significant progress
has been made in sensor technology, there is a need for research focused on sensors capable
of detecting water levels and triggering safety mechanisms in real-time [12].

Despite advancements in vehicle safety and flood mitigation technologies, there remains a
significant gap in comprehensive solutions that address the challenges of vehicle flooding.
Most existing systems focus on either enhancing vehicle resilience to water ingress or
improving communication during emergencies. The integration of buoyancy aids, such as
airbags, with indicating systems that provide real-time alerts has not been extensively
researched. The review of existing literature reveals a critical need for innovative solutions
to enhance vehicle safety during floods. The proposed anti-sinking airbag and indicating
system aims to fill this gap by providing a dual-function approach that prevents vehicles
from submerging and facilitates timely emergency response. By leveraging advancements
in airbag technology, sensor integration, and communication systems, this research seeks to
contribute to the development of more resilient and safer vehicles capable of withstanding
the challenges posed by flooding events.

III Proposed Methodology:

The anti-sinking air bag and indicating system is a systematic integration of electronic
sensors, associated signal conditioning circuits, electronic controller unit, an airbag inflation
system and an indicating unit. Figure 1 depicts the complete system showing the vital parts
of it.

263
International Journal of Research and Analysis in Science and Engineering

Figure 1: Block diagram of the system

The system consists of water level sensor, tilt sensor and impact sensor. Water level sensor
is essential for measuring the water level around the vehicle. Tilt sensor provides the
information about the tilt of a vehicle during flood conditions. Impact sensor provides the
information regarding any accidents that may occur due to flood situations. These three-
sensor data are collected by the control unit for data analysis. Based on these sensor data,
the control unit will send the necessary information to indicating system and the air bag
systems. The control unit is the heart of this system. After receiving the real time data from
the sensors, the control unit processes and analyzes the data. Also, it decides when to inflate
the airbags. The air bags are made up of nylon 66 so that it can provide buoyancy to the
vehicle. Gas generators are also necessary to inflate the air bags quickly in the flood
situation. Along with real time data processing, it has to manage GPS and wireless
communication for sending alerts like emergency service alerts and occupant alerts.
Occupant alerts are in the form of visual and auditory warnings. Also in emergency
situations, the indicating system will provide the location and status information to
emergency responders for the quick services.

Figure 2: Flow chart of measurement and control

264
 Anti-Sinking Airbag and Indicating System for Vehicle during Flood

The structure and workflow of the system highlighting the integration and interaction
between various subsystems to ensure vehicle safety during floods are shown in this block
diagram.

The flow chart for the system is shown in Figure 2. Initially the sensors used in the system
will measure the data. A water level sensor is a device designed to detect the level of water
in a given environment and provide an output signal indicating the water level. It can be
used in various applications, including flood detection systems, industrial processes, and
household appliances. The sensor monitors the presence and level of water. It does so using
various detection mechanisms such as conductivity, pressure, optical, and ultrasonic
methods. The sensor converts the detected water level into an electrical signal. The sensor
sends the processed signal to a control unit and display system, which can trigger actions
like alerts, pump activation and in this system deploying the air bags.

A tilt sensor is a device that measures the tilt or inclination of an object with respect to
gravity. It detects the orientation and motion of an object and converts the tilt into an
electrical signal that can be read by a microcontroller or processing unit. Tilt sensors are
widely used in various applications, including automotive safety systems, mobile devices,
and industrial machinery. Tilt sensors typically use accelerometers, liquid-filled tubes, or
mercury switches to detect changes in orientation. The sensor converts the detected tilt into
an electrical signal. The sensor sends the processed signal to a control unit, which can trigger
actions like alerts or adjustments.

An impact sensor is a device designed to detect and measure sudden forces or shocks exerted
on an object. These sensors are crucial in various applications, including automotive safety
systems, industrial machinery, and consumer electronics, where they help in detecting
collisions, falls, or any sudden impacts. Sensing Mechanism: Impact sensors detect sudden
changes in force or acceleration. They can use various technologies such as piezoelectric
materials, accelerometers, or strain gauges. The sensor converts the detected impact into an
electrical signal. The sensor sends the processed signal to a control unit, which can trigger
actions such as alerts, system shutdowns, or safety mechanisms.

These sensors are crucial for ensuring vehicle safety, especially in dynamic environments
such as during a flood. Their integration into safety systems helps detect potential hazards
and trigger necessary actions to protect the vehicle and its occupants. The control unit is the
microcontroller which can take decisions based on the real time data received from sensors.
A threshold values are defined and if the measured sensor values are more than the
threshold, the control unit will perform two tasks: it sends the emergency alert messages
using GPS and GSM technologies and also the activation signal is provided to air bag
system to deploy air bags.

Using these, the system is able to provide safety to the vehicle and occupants.

IV Discussion and Conclusion:

Flooding is a significant issue worldwide, particularly during the monsoon season in India.
Flood risks affect both human life and vehicles, necessitating proper management of parked
vehicles in flood-prone areas. To address this, an innovative system has been proposed to
265
International Journal of Research and Analysis in Science and Engineering

enhance vehicle safety during floods, which includes an anti-sinking airbag and an
indicating system. Advanced sensors embedded in the vehicle bottom detect rising water
level. An algorithm in the electronic controller ensures rapid and effective airbag inflation.
When floodwater reaches critical levels, airbag is inflated to buoy the vehicle, preventing it
from submerging. The system has multiple advantages: Ensures enhanced safety of vehicle
occupants, real-time updates and alerts to improve the effectiveness of emergency services
and minimizes flood-related vehicle damage.

Designing and developing an intelligent system to buoy the vehicles during floods is a
crucial application that integrates sensors, electronic systems, and airbags. This innovative
system aims to provide significant technological support for rescuing vehicles and alerting
rescue management personnel, thereby enhancing rescue operations during floods. The
proposed dual-function system, comprising an anti-sinking airbag and an indicating system,
offers a significant improvement in vehicle safety during floods. By ensuring rapid response
and effective communication, this system not only protects vehicle occupants but also
contributes to more efficient emergency management during flood events.

References:

1. Ahmed, Mozumdar A., Katharine Haynes, and Mel Taylor. "Vehicle‐related flood
fatalities in Australia, 2001–2017." Journal of flood risk management 13.3 (2020):
e12616.
2. Wang, Na, et al. "A dynamic, convenient and accurate method for assessing the flood
risk of people and vehicle." Science of the total environment 797 (2021): 149036.
3. Wang, Na, et al. "A dynamic, convenient and accurate method for assessing the flood
risk of people and vehicle." Science of the total environment 797 (2021): 149036.
4. Bocanegra, Ricardo A., Francisco J. Vallés‐Morán, and Félix Francés. "Review and
analysis of vehicle stability models during floods and proposal for future
improvements." Journal of flood risk management 13 (2020): e12551.
5. Choo, Kyung-Su, Dong-Ho Kang, and Byung-Sik Kim. "Impact assessment of urban
flood on traffic disruption using rainfall–depth–vehicle speed relationship." Water 12.4
(2020): 926.
6. Milanesi, Luca, and Marco Pilotti. "A conceptual model of vehicles stability in flood
flows." Journal of Hydraulic Research 58.4 (2020): 701-708.
7. Park, Somin, et al. "Computer vision–based estimation of flood depth in flooded-
vehicle images." Journal of Computing in Civil Engineering 35.2 (2021): 04020072.
8. Diakakis, Michalis. "Characteristics of infrastructure and surrounding geo-
environmental circumstances involved in fatal incidents caused by flash flooding:
evidence from Greece." Water 14.5 (2022): 746.
9. Qi, Wenchao, et al. "A review on applications of urban flood models in flood mitigation
strategies." Natural Hazards 108 (2021): 31-62.
10. Botonis, Olivia K., et al. "Wearable airbag technology and machine learned models to
mitigate falls after stroke." Journal of neuro engineering and rehabilitation 19.1 (2022):
60.
11. Campolo, Claudia, et al. "Modeling prioritized broadcasting in multichannel vehicular
networks." IEEE Transactions on Vehicular Technology 61.2 (2011): 687-701.
12. Wang, Zhendong, and Huamao Xie. "Wireless sensor network deployment of 3D
surface based on enhanced grey wolf optimizer." IEEE Access 8 (2020): 57229-57251.
266
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

28. Combined Effect of Noise Reduction and

Multiband Frequency Compression for
Improving Speech Perception in Monaural
Hearing Aids on Source Localization
Jyoti M. Katagi
Dept. of Electronics & Communication Engineering,
Biluru Gurubasava Mahaswamiji Institute of Technology,
Mudhol, Karnataka, India.
Pandurangarao N. Kulkarni
Dept. of Electronics & Communication Engineering,
Basaveshwar Engineering College,
Bagalkote, Karnataka, India.

Abstract:

People with sensorineural hearing loss have wider auditory filters. The wider auditory
filters have relatively smooth spectrum representations. This induces spectral masking,
which impairs hearing-impaired people's ability to understand speech. The speech becomes
less understandable when there is background noise, too. Therefore, it is important to select
the best hearing aid algorithms like frequency lowering, frequency transposition, and
frequency compression to minimize the effects of spectral masking. In order to enhance
speech perception, it is therefore imperative to use noise reduction techniques in
conjunction with hearing aid algorithms, but typically, these parts are created and
evaluated separately. Therefore, the goal of the current study is to evaluate the combined
effect of multiband frequency compression and noise reduction techniques on sound source
localization for improving speech perception in monaural hearing aids. In the present work,
we have investigated the impact of this approach on source localization with a compression
factor of 0.6. The listening tests conducted for 7 different azimuth angles (-90⁰, -60⁰, -30⁰, 0⁰,
30⁰, 60⁰, and 90⁰) on 6 listeners with normal hearing under various signal to noise ratio
(SNR) situations and on 6 listeners with mild sensorineural hearing loss (SNHL) showed
that there is no detrimental effect on localization.

Keywords:

Hearing aids, multi-band frequency compression, Noise reduction, Sensorineural hearing

loss, Source localization.

267
International Journal of Research and Analysis in Science and Engineering

I Introduction:

Humans deal with localization of sound on every day: when crossing a roadway, it is
necessary to know if an automobile will be coming or not, this is done either through
visualization or with the sound source localization. A person must also identify the direction
of sound source when someone shouts their name. To do all this brain is required to pinpoint
the location of an event's source position hundreds of times every day. But how humans are
doing it? The fact that people have two ears is the essential factor. This feature, together
with the unique structure of the outer ear (pinna), contributes to humans' incredible ability
to locate sound sources is discussed in thesis [1]. Numerous technical applications, such as
locating an active speaker or enhancing the signal to noise ratio in hearing aids, depend on
localization.

Intermural time difference (ITD), intermural level difference (ILD), and spectral cues are
three acoustic cues that must be perceptually integrated in order to localize a sound source
[2] - [6]. In order to locate sources in the horizontal plane, ITD and ILD are the most
essential signals. ILD is significant at higher frequencies, whereas ITD is significant at
lower frequencies, based on the "duplex" concept of binaural localization [7]. When locating
in the vertical plane and differentiating between the rear and the front, spectral hints of the
high frequency (> 5 kHz) signals produced by sound diffraction by the pinna are important.
ITD varies with the distance between the two ears, ranging from a minimum value of 0 for
a sound originating from directly ahead to a value of around 690 μs for a sound coming
from a source directly across from one ear. ILD ranges from 0 to 20 dB [8], [9].

Over the years, many attempts have been made to determine the relationship between the
location of a sound source in space and the sound pressure that source generates at a
listener's eardrums. Traditionally, this link has been shown using the head related transfer
function, or HRTF [10], [11].

Compared to individuals who can hear properly, hearing-impaired people have a wider
auditory filter. Frequency selectivity decreases due to increased masking effects. Speech
perception is often impaired by sensorineural hearing loss due to increased hearing levels,
intensity recruitment, a lower dynamic range, and raised temporal and spectral masking
[12]. Automated gain control, frequency selective gain, and multichannel dynamic range
compression with customizable release time, number of channels, attack time, and
compression ratios are features found in many hearing aids [13], [14]. In order to further
enhance speech perception, many techniques have been suggested to lessen the effects of
maximum intra speech spectral masking brought on by broadening of auditory filters [15] -
[17]. By studying all these techniques in this work we have come up with combined effect
of noise reduction and multiband frequency compression for improving speech perception
in monaural hearing aids on source localization.

The frequency compression technique is built on the basis of auditory critical bands [18] -
[20]. With this technique, the voice signal was pushed into the middle of every significant
band along the frequency axis. FFT was calculated for very frame after the input voice was
separated into segments using a Hamming window. It was compressed in the range of 0.1
and 0.9. After piecewise frequency compression, the magnitude spectrum and the original
phase spectrum were combined.
268
 Combined Effect of Noise Reduction and Multiband Frequency Compression for Improving…

The new voice signal was created using the overlap-add technique. Participants with hearing
impairments took part in listening tests fifty vowel-consonant-vowel (VCV) syllables were
used as the test material from a male speaker. With compression between 0.2 and 0.4, the
recognition percentage went from 35.4% to 38.3% for the unprocessed set to the greatest
performance.

A comparison of horizontal localization and speech perception in noise with and without
digital noise reduction (DNR) activation in hearing aids with and without an ear to ear
synchronization is given in [21]. Participants were 25 listeners with mild to moderate
bilateral sensorineural hearing loss, ranging in age from 18 to 55 years. A root-mean-square
degree of inaccuracy was used to assess each participant's ability to locate horizontal sound
sources. The signal-to-noise ratio needed to get a 50% recognition score (SNR-50) for voice
recognition in the presence of speech babble noise was calculated. Additionally, SNR-50
was assessed using noise sources coming from four distinct angles, and it was recorded
under four assisted scenarios both with and without the independent activation of wireless
links and DNR. According to the results of the current study, hearing aids with wireless
synchronization and DNR turned on performed better across all of the metrics.
Nevertheless, depending on the direction of noise and speech, the increase in scores may or
may not be advantageous to the listener. The subsequent sections present the proposed work,
tests, results and conclusion.

II Proposed Method:

A coordinate system, as shown in Figure 1, must be provided in order to describe

localization in the three-dimensional space.

Figure 1: Coordinate system relative to the head with azimuth θ and elevation φ

Azimuth (θ), which also refers to the horizontal plane that determines the left-to-right
direction, is a term used to describe how the sound source is positioned in relation to the
head. Elevation (ϕ) refers to the vertical surface that indicates the direction of upward and
downward motion. The horizontal surface is represented by the x1 and x2-axes in Figure 1,
while the vertical surface is represented by the x2 and x3-axes. Although sound enters both
ears, the brain distinguishes between information obtained from binaural signals and
information obtained from monaural signals. Information concerning the interaural time
difference (ITD) and interaural level (intensity) difference (ILD, IID) is sent by the signals
269
International Journal of Research and Analysis in Science and Engineering

that reach the left and right ears. This indicates that the incident sound wave reaches our
ears at various times and volumes. One ear receives the sound before the other as long as
the source is not directly in front of the head. Both ears signals are used to retrieve this
information. The term "binaural cues" is used to describe them. Since ILD and ITD are the
same for the left and right ears, sound sources that come directly in front of or behind a
person do not provide them. Monaural signals are significant in this situation. The spectral
structuring of the incident sound waves by the head, shoulders, torso, and, most importantly,
the pinna, is represented by monaural cues, which are retrieved from the signal of one ear.

Elevation, azimuth and time all influence the head related impulse response 'h'. HRIR is
sampled in the data files both temporally and spatially. The discrete indices naz, nel, and
nt are used to specify azimuth, elevation, and time, respectively. A 3D array having
dimensions of (25*50*200) is an HRIR h (naz, nel, nt). HRIR values are given for 25
different azimuths, 50 different elevations and 200 instants in time [22]. The azimuth is the
angle between a vector to the sound source and the midsagittal or vertical median plane and
varies from -90° to +90°. The elevation is the angle from the horizontal plane to the
projection of the source into the midsagittal plane and varies from -90° and +270°. The
azimuth index naz is related to the azimuth angle θ as follows

naz θ naz θ naz θ naz θ naz θ

1 -80⁰ 6 -35⁰ 11 -10⁰ 16 15⁰ 21 40⁰
2 -65⁰ 7 -30⁰ 12 -5⁰ 17 20⁰ 22 45⁰
3 -55⁰ 8 -25⁰ 13 0⁰ 18 25⁰ 23 55⁰
4 -45⁰ 9 -20⁰ 14 5⁰ 19 30⁰ 24 65⁰

In MATLAB, the azimuth angle corresponding to naz is the naz-th element of the vector

Azimuths= [-80, -65, -55, -45: 5: 45, 55, 65, 80].

Both elevation and time are uniformly sampled. Elevations range from -45⁰ to 230.625⁰ in
steps of 5.625⁰. In MATLAB, the elevation angle corresponding to nel is the nel-th element
of the vector

Elevations= -45 + 5.625 *(0:49)

In the inter aural polar coordinate system, Table 1 shows the range of azimuth and elevation
angle values for various places.

Table I: Azimuth and Elevation Directions in 3d Space

Azimuth Elevation Direction in 3D space

0⁰ 0⁰ Ahead
0⁰ 90⁰ Overhead
0⁰ 180⁰ Behind
0⁰ 270⁰ Below
270
 Combined Effect of Noise Reduction and Multiband Frequency Compression for Improving…

Azimuth Elevation Direction in 3D space

90⁰ 0⁰ To the right
-90⁰ 0⁰ To the left

III Tests and Results:

The present work aims to evaluate the effects of multiband frequency compression and
cascaded noise reduction on source localization in order to enhance speech perception for
monaural hearing aids. During the experiment, two HRTFs were used to produce spatial
sounds. The public domain CIPIC HRTF database provides HRTFs for various azimuth and
elevation configurations, as well as a detailed explanation of the method used to quantify
HRTFs and anthropometric features [23]. The HRTFs from this database for Subject 3, a
KEMAR manikin participant, were used in the current experiment. They were for elevation
angle of 0⁰ and the frontal azimuth angles ranging from -90⁰ (left) to +90⁰ (right).

Six individuals who had normal hearing in the face of broad band masking noise and six
individuals with mild sensorineural loss underwent hearing tests to study source
localization. Broad-band random noise was used as a masker to the processed stimuli for
participants with normal hearing. It was administered for a brief (10 ms) period of time at
six SNR values: ∞, 6, 3, 0, -3, and -6dB. Without utilizing broad-band masking noise,
hearing-impaired subjects were evaluated. Every time a test was administered, participants
were given the option to choose their level of comfort with the binaurally presented sounds.

The aim of present work was to compare direction identification outcomes in processed and
unprocessed conditions. The initial hearing tests were performed on 6 normal subjects while
they were subjected to a broadband masking noise. Six SNR values were utilized to induce
the noise: ∞ (no noise), 6, 3, 0, -3, and -6dB. Six subjects with modest bilateral sensorineural
loss underwent the second round of testing without the use of masking noise. In both
experiments, sounds were processed using HRTFs with 0⁰ elevation and 0⁰, ±30⁰, ±60⁰, and
±90⁰ azimuth angles. The participant was presented with a chart that detailed these
directions, as shown in Fig.2. The stimuli that are processed for the various angles are
delivered in a random sequence and 5 times every angle is repeated. One of these seven
angles was recognized by the person as the direction of the source. Each perceived angle's
mean was computed using the responses as columns in a stimulus response matrix.

Broad band noise and the sound of breaking glass as an ambient sound served as the test
stimuli in the experiment with individuals who had normal hearing. All that was used as the
test stimuli for the masking noise experiment was the sound of breaking glass. Therefore,
each subject received a total of 210 presentations with masking noise (7 angles x 5
repetitions x 6 SNR values) and 35 presentations with no noise (7 angles x 5 repetitions).
The stimulus-response matrix for the sound of breaking glass is depicted in Table 2 with
responses from all six normal hearing participants combined. In an experiment with hearing
impaired subjects, the sound of breaking glass served as one of the test stimuli. Each person
received a total of 35 presentations (7 angles x 5 repetitions). Table 3 displays the stimulus-
response matrix for the sound of breaking glass with the responses from all six hearing-
impaired participants combined.

271
International Journal of Research and Analysis in Science and Engineering

Figure 2: Azimuth Perception Test Reference Chart

Using wiener filters as a noise reduction approach, the suggested scheme processes the
speech data to get the angle determination score (%) that is displayed in Figure 3. Six
individuals with hearing impairments took part in the experiment. Based on these figures,
we can see that there were no negative impacts on source localization according to the
subjective evaluation of seven different azimuth angles conducted on six listeners with
hearing impairment.

A. Graphical Analysis:

100.00%

80.00%

60.00%

40.00%

20.00%

0.00%
-90⁰ -60⁰ -30⁰ 0⁰ 30⁰ 60⁰ 90⁰
Unprocessed Processed

Figure 3: Angle determination score (%) in unprocessed & processed scenarios for 6
Hearing Impaired participants

B. Spectrographic Analysis:

Figures 4 and 5 below depict the hearing-impaired individuals left ear, right ear and wide
band spectrogram of the unprocessed and processed glass breaking speech signal at a 90-
degree angle. Figures 6 and 7 below depict normal hearing persons left ear, right ear and
wide band spectrogram of the unprocessed and processed glass breaking speech signal
spectra at a 90-degree angle and -6 dB. Sound spectrograms that have been processed
indicate that background noise has been mostly eliminated and that the harmonic structure
is unaffected by speech compression.
272
 Combined Effect of Noise Reduction and Multiband Frequency Compression for Improving…

In case of normal hearing people unprocessed speech is the input speech with different
SNR’s and processed speech is the output from cascaded structure of noise reduction
followed multiband frequency compression. In case of hearing impaired people unprocessed
speech is the clean input speech and processed speech is the output from cascaded structure
of noise reduction followed multiband frequency compression.

For unprocessed and processed speech with noise reduction and multiband frequency
compression at several SNR levels and a compression ratio of 0.6, the average source
direction identification by six participants with normal hearing is shown in Table 2. In
comparison to raw voice at SNR values of ∞ dB, +6 dB, +3 dB, 0 dB, -3 dB, and -6 dB,
respectively, the mean of processed speech values improved to 1.14, 4.14, 6.15, 5, 7.71, and
8.57 in the sound source direction identification. Additionally, it has been found that at
lower SNR values, improvements in the ability to localize sound sources are considerable.
The findings indicate that the individuals were able to perceive the source direction by
utilizing ITD and ILD signals from different bands. For six hearing-impaired participants,
Table 4 compares their percentages of angle recognition scores under unprocessed and
processed situations using the stimulus of glass breaking sound. The percentage of the
average stimulus-response matrix that has not been processed for the six hearing-impaired
participants is 33.3%, 53.3%, 50%, 90%, 23.3%, 60%, and 66.7% at azimuth angles of -90⁰,
-60⁰, -30⁰, 0⁰, 30⁰, 60⁰ and 90⁰ respectively. We can find that the localization performance is
only up to 53.8% from the unprocessed stimulus-response matrix produced by performing
listening tests. The percentage of the average stimulus-response matrix that has been
processed among six hearing-impaired participants is 53.3%, 63.3%, 56.7%, 93.3%, 33.3%,
53.3%, and 70%, respectively, for azimuth angles of -90⁰, -60⁰, -30⁰, 0⁰, 30⁰, 60⁰ and 90⁰.
According to the processed stimulus-response matrix that was acquired from the listening
tests, the localization performance was only up to 60.46%.

The subjective assessment for 7 separate azimuth angles on 6 listeners with normal hearing
under different signal-to-noise ratio circumstances and 6 listeners with hearing impairment
found no detrimental effects on source localization

Figure 4: Hearing Impaired People Left Ear, Right Ear and Wide Band Spectrogram
of Unprocessed Glass Breaking Speech Signal for 90-Degree Angle

273
International Journal of Research and Analysis in Science and Engineering

Figure 5: Hearing Impaired People Left ear, Right ear and Wide Band Spectrogram
of processed glass breaking speech signal for 90-degree angle

Figure 6: Normal hearing people Left ear, Right ear and Wide band Spectrogram of
unprocessed glass breaking speech signal for -6 dB at 90 degree angle

Figure 7: Normal hearing people Left ear, Right ear and Wide Band Spectrogram of
processed glass breaking speech signal for -6 dB at 90-degree angle

274
 Combined Effect of Noise Reduction and Multiband Frequency Compression for Improving…

Table II: Average Source Localization Values for Processed and Unprocessed Speech
for The Six Normal Hearing Individuals at Various Snr Levels with The Sound of
Breaking Glass

SNR (dB)
∞ 6 3 0 -3 -6
Angle Un Proc Un Proc Un Pr Un Proc Un Proc Un Proc
(deg) proc esse proc esse proc oce proc esse proc esse proc esse
esse d esse d esse ss esse d esse d esse d
d d d ed d d d
⁰
-90 20 22 21 26 15 25 20 26 17 29 14 26
⁰
-60 16 17 17 22 14 24 15 21 14 24 12 21
⁰
-30 14 16 13 17 17 20 14 18 12 20 10 19
⁰
0 30 30 30 30 30 30 30 30 30 30 30 30
⁰
30 15 16 14 20 13 21 12 19 14 19 11 20
⁰
60 16 17 20 23 15 20 15 23 13 24 12 22
90⁰ 21 22 21 27 20 27 21 25 21 29 16 27
Mean 18.8 19.4 23.5 17.7 23. 18.1 23.1 17.2 23.5
20 25 15
6 3 7 1 86 4 4 9 7
Impro
veme 1.14 4.14 6.15 5 7.71 8.57
nt

Table III: Source Localization Scores for Presentation Angle Versus Perceived
Angle in Hearing-Impaired Individuals. There Are a Total of 30 Presentations (5
Presentations X 6 Subjects) For Each Angle. Glass Breaking Sound Is the Test
Material

Presented Unprocessed Speech Processed Speech

Azimuth Perceived angle (deg.) Perceived angle (deg.)
angle(deg.)
- - - - - -
0⁰ 30⁰ 60⁰ 90⁰ 0⁰ 30⁰ 60⁰ 90⁰
90⁰ 60⁰ 30⁰ 90⁰ 60⁰ 30⁰
-90⁰ 10 12 8 0 0 0 0 16 10 4 0 0 0 0
-60⁰ 5 16 9 0 0 0 0 7 19 4 0 0 0 0
-30⁰ 6 9 15 0 0 0 0 3 10 17 0 0 0 0
0⁰ 0 0 2 27 1 0 0 0 0 2 28 0 0 0
30⁰ 0 0 0 0 7 15 8 0 0 0 3 10 13 4
60⁰ 0 0 0 0 3 18 9 0 0 0 0 6 16 8
90⁰ 0 0 0 0 0 10 20 0 0 0 0 1 8 21

275
International Journal of Research and Analysis in Science and Engineering

Table IV: Angle Determination Score (%) In Unprocessed and Processed Scenarios
for Six Hearing-Impaired Participants

Prese Unprocessed Speech Processed Speech

nted Perceived angle (deg.) Perceived angle (deg.)
Azim
uth ⁰ ⁰ ⁰ ⁰
angle -90 -60 -30 0 30⁰ 60⁰ 90⁰ -90⁰ -60⁰ -30⁰ 0⁰ 30⁰ 60⁰ 90⁰
(deg.)
33. 40 26. 0 0% 0% 0% 53.3 33. 13. 0% 0% 0% 0%
-90⁰
3% % 7% % % 3% 3%
16. 53. 30 0 0% 0% 0% 23.3 63. 13. 0% 0% 0% 0%
-60⁰
7% 3% % % % 3% 3%
20 30 50 0 0% 0% 0% 10% 33. 56. 0% 0% 0% 0%
-30⁰
% % % % 3% 7%
0% 0% 6.6 90 3.3 0% 0% 0% 0% 6.6 93. 0% 0% 0%
0⁰
7% % 3% 7% 3%
0% 0% 0% 0 23. 50 26. 0% 0% 0% 10 33. 43. 13.
30⁰
% 3% % 7% % 3% 3% 3%
0% 0% 0% 0 10 60 30 0% 0% 0% 0% 20 53. 26.
60⁰
% % % % % 3% 7%
0% 0% 0% 0 0% 33. 66. 0% 0% 0% 0% 3.3 26. 70
90⁰
% 3% 7% 3% 7% %

IV Conclusion:

Based on previous studies and the findings from the current investigation, we can conclude
that the issue of noise suppression, localization, and concerns particular to sensorineural
hearing loss peoples, such as lower dynamic range, spectral and temporal masking, have
received a substantial amount of attention for hearing aids. So, in addition to addressing the
negative consequences of SNHL, a method that lowers background noise should also be
included. In the current work, we have taken into account a cascaded structural mechanism
that takes into account both the noise suppression and sensorineural hearing loss issues,
namely a decreased dynamic range and spectral masking to study their impact on source
localization. Comparing our previous work [24] on effect of multiband frequency
compression for enhancing speech perception in monaural hearing aids on source
localization with the current work (Noise minimization and Multi-band frequency
compression), there is a little change over the source localization score of about 5.24% for
processed speech in the latter one. Therefore, cascading structure of noise suppression and
multiple-band frequency compression plays important role in improving speech perception
in monaural hearing aids. If this method is applied to the analysis of speech signals in
hearing aids, people with moderate sensorineural loss may be able to hear speech more
clearly. To determine how effectively this strategy enhances speech perception and how
noise reduction techniques impact source localization has to be evaluated with a larger
276
 Combined Effect of Noise Reduction and Multiband Frequency Compression for Improving…

number of subjects and a variety of test materials in the future work.

References:

1. Anna Katharina Fuchs, “Monaural Sound Localization”, Diploma Thesis, Signal

Processing and Speech Communication Lab, Graz University of Technology, Austria,
2010.
2. Stevens S. S., & Newman E. B., “The localization of actual sources of sound”,
American Journal of Psychology. 48, 297 - 306, 1936.
3. Hartmann W. M., “How we localize sound”, Physics Today. 11, 24 - 29, 1999.
4. Chung W., Carlile S., & Leong P., “A performance adequate computational model for
auditory localization”, Journal of the Acoustical Society of America 107, 432 - 445,
2000.
5. Langedijk E. H. A., & Bronkhorst A. W., “Contribution of spectral cues for human
sound localization”, Journal of the Acoustical Society of America 112, 1583 - 1596,
2002.
6. Best V., Carlile S., Jin C., & Van Schaik A., “The role of high frequencies in source
localization”, Journal of the Acoustical Society of America 118, 353 - 363, 2005.
7. Simon H. J., “Bilateral amplification and sound localization: then and now”, Journal of
Rehabilitation Research and Development 42, 117 - 132, 2005.
8. Middlebrooks J. C., “Narrow-band sound localization related to external ear acoustics”,
Journal of the Acoustical Society of America 92, 2607 - 2624, 1992.
9. Van den Bogaert T., Klasen T. J., Moonen M., Van Deun L., & Wouters J., “Horizontal
localization with bilateral hearing aids: without is better than with”, Journal of the
Acoustical Society of America 119, 515 - 526, 2006.
10. Douglas S. Brungart, & Brian D. Simpson, “Auditory Localization of nearby sources in
a Virtual Audio Display”, Proceedings of the IEEE Workshop on the Applications of
Signal Processing to Audio and Acoustics, New Platz, NY, USA, Oct.2001.
11. Brungart D. S., & Rabinowitz W. R., “Auditory localization of nearby sources. Head
related transfer functions”, Journal of the Acoustical Society of America. 106, 1465 -
1479, 1999.
12. Moore B.C.J., An introduction to the Psychology of Hearing. Academic, London, 1997.
13. J. M. Pickett, “The Acoustics of Speech Communication: Fundamentals, Speech
Perception Theory, and Technology”, Boston, Massachusetts: Allyn Bacon, pp.289 -
323, 1999.
14. H. Dillon, Hearing Aids. New York: Thieme Medical Publisher, 2001.
15. P. N. Kulkarni, P. C. Pandey, & D. S. Jangamashetti, “Multi-band frequency
compression for improving speech perception by listeners with moderate sensorineural
hearing loss”, Speech Communication, vol.54, No. 3, pp.341 - 350, March 2012.
16. P. N. Kulkarni, P. C. Pandey, & D. S. Jangamashetti, “Multi-band frequency
compression for reducing the effects of spectral masking”, International Journal of
Speech Technology. vol.10, 219 - 227, 2007.
17. P. N. Kulkarni, “Speech processing for reducing the effects of spectral masking in
sensorineural hearing loss”, Ph.D. Thesis, Department of Electrical Engineering, Indian
Institute of Technology Bombay, India, 2010.
18. K. Yasu, K. Kobayashi, K. Shinohara, M. Hishitani, T. Arai, & Y. Murahara,
“Frequency compression of critical band for digital hearing aids”, China-Japan Joint
Conference on Acoustics, pp.159 - 162, 2002.

277
International Journal of Research and Analysis in Science and Engineering

19. T. Arai, K. Yasu, & N. Hodoshima, “Effective speech processing for various impaired
listeners”, Proc. 18th International Congress on Acoustics, pp.1389 - 1392, 2004.
20. S. Sakamoto, K. Goto, M. Tateno, & K. Kaga, “Frequency compression hearing aid for
severe-to-profound hearing impairment”, Auris Nasus Larynx, 27, pp.327 - 334, 2000.
21. Geetha Chinnaraj, Kishore Tanniru, & Raja Rajan Raveendran, “Speech perception in
noise and localization performance of digital noise reduction algorithm in hearing aids
with ear-to-ear synchronization”, Journal of The All India Institute of Speech and
Hearing, pp.23 - 30, Volume 40, Issue 1, September 2022.
22. “CIPIC database”, http://interface.cipic.ucdavis.edu.
23. V. R. Algazi, R. O. Duda, D. M. Thompson, & C. Avendano, “Then CIPIC HRTF
database”, Proceedings of the 2001 IEEE Workshop on the Applications of Signal
Processing to Audio and Acoustics, New Platz, NY, pp.99 – 102, 2001.
24. Jyoti M. Katagi, & Pandurangarao N. Kulkarni, “Effect of Multi-band Frequency
Compression for Improving Speech Perception in Monaural Hearing Aids on Source
Localization”, IEEE North Karnataka Subsection Flagship International Conference
(NKCon), BLDEA's V. P. Dr. P. G. Halakatti College of Engineering and Technology,
Vijayapur, 20 - 21 Nov 2022.

278
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

29. Data Management in Edge Computing:

Opportunities and Challenges
Piyusha S. Shetgar
Electronics and Telecommunication Engineering,
Walchand Institute of Technology,
Solapur, Maharashtra, India.
Chayalakshmi C. L., Mahabaleshwar S. K.,
Rajani S. Pujar
Electronics and Communication Engineering,
Basaveshwar Engineering College,
Bagalkote, Karnataka, India
Abstract:

In order to handle the enormous volumes of data produced by IoT and IIoT devices, edge
computing has become an essential technological advancement. This review paper explores
recent advancements in data management using edge computing, focusing on efficient data
placement, retrieval, reduction mechanisms, privacy-preserving techniques, and energy-
efficient scheduling. Additionally, the paper identifies existing gaps and challenges, offering
insights into future research directions. Edge computing is a paradigm shift in data
management, bringing computation and storage closer to the source. This article examines
the introduction to edge computing, including its architecture, important advantages,
problems, and real-world applications. The goal is to give a detailed overview of how edge
computing can transform data management processes, improve performance, and
overcome the constraints of standard cloud computing.

Keywords:

Edge computing, data management, data reduction, privacy-preserving, energy efficiency

I Introduction:

The explosion of IoT and IIoT devices has resulted in an unprecedented increase in data
generation. Traditional cloud computing infrastructure is under pressure due to the
exponential expansion of data created by Internet of Things (IoT) devices, mobile
applications, and other digital services. The requirement for real-time data processing, low
latency, and efficient bandwidth utilization has spurred the emergence of edge computing.

279
International Journal of Research and Analysis in Science and Engineering

Traditional cloud computing infrastructures struggle to cope with this data influx, leading
to latency issues, bandwidth constraints, and increased costs. By processing data closer to
its source, edge computing offers a workable solution by lowering latency and speeding up
reaction times.

A. Edge Computing Architecture:

Edge computing architecture involves several steps to ensure efficient processing of data
close to the source. Here are the typical steps involved in designing and implementing an
edge computing architecture:

Figure 1: Edge Computing Architecture

Edge computing architecture comprises several key components:

1. Edge Devices: These include sensors, IoT devices, and other endpoints that generate
data. Examples are smart thermostats, industrial sensors, and mobile devices.
2. Edge Nodes: Intermediate devices such as gateways, routers, and local servers that
process and store data closer to the edge devices. These nodes perform preliminary data
processing, filtering, and aggregation before sending relevant data to cloud servers if
necessary.
3. Edge Data Centers: Smaller-scale data centers that provide additional processing and
storage capabilities. They act as miniaturized versions of central data centers but are
strategically located closer to the edge devices to reduce latency.
4. Cloud Data Centers: Centralized data centers that handle tasks that cannot be
efficiently managed at the edge. They offer substantial computational power and storage
capacity, supporting extensive data analysis, machine learning, and long-term storage.

280
 Data Management in Edge Computing: Opportunities and Challenges

B. Edge Computing Advantages Versus Cloud Computing:

Figure 2: Edge Computing Advantages Versus Cloud Computing

1. Closeness to Data Source: By processing data locally, edge computing eliminates the
need for data to travel vast distances to cloud data centers. Because data doesn't have to
travel across the entire network to reach a central server, this close proximity improves
speed and efficiency.
2. Real-Time Processing: Applications like driverless cars, industrial automation, and
healthcare depend on real-time data processing and decision-making, which edge
computing enables. Timely responses and actions are ensured via immediate data
processing at the edge.
3. Lower Latency: Compared to cloud computing, edge computing offers faster response
times by drastically lowering latency by shortening the distance data must travel. For
latency-sensitive applications, where even little delays can have major repercussions,
this is essential.
4. Cost Savings: By processing and storing data locally, edge computing can save
expenses related to data transfer and cloud storage. Lowering the amount of data
transferred to the cloud reduces the need for pricey cloud storage options as well as the
cost of transmission.
5. Improved Bandwidth Utilization: By handling data processing at the edge, the strain
on network bandwidth is reduced, leading to more efficient use of network resources.
This is particularly important in environments with limited bandwidth or high data
traffic.
6. Improved Security: By processing data locally, hazardous information is not as likely
to be exposed when being transferred to and from cloud servers. The chance of illegal
access and data breaches is reduced when data processing is done locally.

281
International Journal of Research and Analysis in Science and Engineering

II Literature Survey:

A review of recent studies on data management in edge computing is presented in this

section, with highlighting efficient data placement and retrieval, prediction-based data
reduction, privacy-preserving methods, energy-efficient scheduling, Edge-Cloud Solutions
for Big Data Analysis, Distributed Deep Learning and Offloading, Edge AI for Real-Time
Processing, Deep Reinforcement Learning for Resource Allocation, Efficient Data
Compression Techniques, and Efficient Data offloading etc. It also addresses current gaps
and challenges in the field mentioned.

A. Edge-Cloud Solutions for Big Data Analysis:

Hybrid solutions for distributed machine learning and large data analysis are made possible
by the combination of edge and cloud computing. Edge-cloud systems, which combine the
low-latency benefits of edge computing with the processing capacity of the cloud, were
covered by Loris Belcastro et al. [1]. The focus of their study is on the smooth
synchronization of data across cloud and edge environments.

Gaps and Challenges:

• Data Synchronization: Guaranteeing effective and smooth data synchronization

between cloud and edge systems.
• Latency Issues: Minimizing latency in hybrid edge-cloud environments.
• Cost Management: Managing costs associated with data transfer and storage between
edge and cloud.

B. Edge AI for Real-Time Processing:

Real-time processing is made possible by Edge AI by allowing machine learning models to

be deployed directly on edge devices. The use of AI models on edge devices, which greatly
lowers latency and enhances real-time decision-making capabilities, is covered by Louis
Frank [2]. Applications that demand quick reactions, like industrial automation and
autonomous driving, benefit greatly from this strategy.

Gaps and Challenges:

• Model Complexity: Deploying complex AI models on resource-constrained edge

devices is challenging.
• Real-Time Processing: Ensuring real-time processing while maintaining model
accuracy and performance.
• Energy Efficiency: Balancing the computational load with energy consumption for AI
processing on edge devices.

C. Prediction-Based Data Reduction:

Techniques for reducing data are essential for handling the enormous volumes of data that
IIoT devices create. An effective edge data management paradigm using prediction-based
282
 Data Management in Edge Computing: Opportunities and Challenges

data reduction methods was presented by Lei Yang et al. [4]. By predicting data trends and
removing redundant data, this framework uses machine learning techniques to drastically
lower storage needs and transmission expenses. Through selective edge processing of
critical data, this strategy improves the overall IIoT system efficiency.

Gaps and Challenges:

• Accuracy of Predictions: The accuracy of predictive models can vary, leading to

potential data loss or inaccuracies.
• Resource Constraints: Implementing sophisticated machine learning algorithms on
resource-constrained edge devices is challenging.
• Real-time Processing: Achieving real-time data reduction while maintaining
prediction accuracy is complex.

D. Privacy-Preserving Techniques:

Edge computing raises serious privacy issues, especially for sensitive applications like
healthcare. Using edge computing, Lingbin Meng and Daofeng Li created a privacy-
preserving method for smart healthcare systems [5]. By processing critical patient data
locally at the edge, their approach minimizes exposure to external threats. Furthermore, a
federated learning framework for privacy-preserving big data analysis in Internet of Medical
Things (IoMT) was proposed by Akarsh K. Nair et al. This system ensures data privacy and
security by enabling many edge devices to train machine learning models cooperatively
without exchanging raw data [6].

Gaps and Challenges:

• Data Security: Ensuring robust data security measures at the edge is critical.
• Communication Overhead: Federated learning can introduce significant
communication overhead.
• Scalability: Scaling privacy-preserving techniques to large networks of edge devices is
challenging.

E. Energy-Efficient Scheduling:

Energy efficiency is a critical factor in the design of edge computing systems. Jing Liu et
al. explored intelligent energy-efficient scheduling techniques using ant colony
optimization for heterogeneous edge computing environments [7]. Their approach
dynamically adjusts the scheduling of computational tasks based on real-time energy
consumption metrics. Similarly, the study by Quy Vu Khanh et al. on sustainable smart
cities proposes an efficient edge computing management mechanism, highlighting
strategies for energy-efficient resource allocation and management [9].

Gaps and Challenges:

• Resource Allocation: Dynamic and efficient resource allocation is complex.

283
International Journal of Research and Analysis in Science and Engineering

• Energy Consumption: Balancing energy consumption with performance remains a

challenge.
• Heterogeneous Environments: Managing heterogeneous edge environments with
diverse devices and capabilities is difficult.

F. Deep Reinforcement Learning for Resource Allocation:

Tongke Cui et al. investigated deep reinforcement learning-based resource allocation for
content distribution in IoT-edge-cloud computing environments [8]. Their approach
leverages reinforcement learning to dynamically allocate resources, ensuring optimal
performance and resource utilization. This method is particularly effective in complex and
dynamic environments where traditional static allocation strategies fall short. Nain et al.
present a comprehensive study on resource optimization in edge computing integrated with
Software-Defined Networking (SDN) [3]. This integration aims to enhance the efficiency,
flexibility, and manageability of edge computing environments by leveraging SDN's
centralized control and programmability features.

Gaps and Challenges:

• Algorithm Complexity: Managing the complexity of deep reinforcement learning

algorithms.
• Real-Time Adaptation: Ensuring real-time adaptation of resource allocation
strategies.
• Scalability: Scaling reinforcement learning-based approaches to large networks of edge
devices.

G. Efficient Data Compression Techniques:

Another essential component of effectively managing data on edge devices is data

compression. Nerea Gómez Larrakoetxea et al. suggested using data compression
approaches to enable effective machine learning on edge computing [10] Their work
focuses on data compression to make better use of edge computing capabilities by lowering
transmission costs and storage requirements without sacrificing much information.

Gaps and Challenges:

• Compression Algorithms: Developing efficient compression algorithms that maintain

data integrity.
• Resource Constraints: Implementing compression techniques on resource-limited
edge devices.
• Latency: Minimizing the latency introduced by data compression and decompression
processes.

H. Distributed Deep Learning and Offloading:

Computational offloading and distributed deep learning are essential for improving edge
computing networks' capabilities. Distributed deep learning-based offloading strategies for
mobile edge computing networks were studied by Liang Huang et al. Their method divides
284
 Data Management in Edge Computing: Opportunities and Challenges

the processing burden among several edge devices [12]. Deep reinforcement learning-based
resource allocation for content distribution in IoT-edge-cloud systems was investigated by
Tongke Cui et al. [11].

Gaps and Challenges:

• Workload Distribution: Efficiently distributing workloads across multiple edge

devices.
• Network Stability: Ensuring network stability and reliability during offloading.
• Algorithm Complexity: Managing the complexity of distributed deep learning
algorithms.

I. Efficient Data Offloading:

Mingye Li et al. explored efficient data offloading using a Markovian decision process in
edge computing. Their approach optimizes the decision-making process for offloading data
from edge to cloud, considering factors such as network conditions, computational load,
and energy consumption [10]. This ensures efficient utilization of resources while
maintaining system performance.

Gaps and Challenges:

• Decision Models: Developing accurate and efficient decision models for data
offloading.
• Network Stability: Ensuring network stability and reliability during offloading.
• Resource Management: Balancing the load between edge and cloud resources
effectively.

J. Efficient Data Placement and Retrieval:

Optimizing edge computing performance requires efficient data insertion and retrieval. A
system for effective data insertion and retrieval in edge contexts was presented by Junjie
Xie et al. By distributing data among several edge nodes in a deliberate manner, their
method reduces latency in data access and increases data availability [13]. The system
ensures efficient data distribution and fast retrieval by using caching, replication, and
partitioning algorithms.

Gaps and Challenges:

• Dynamic Environments: Existing models often assume static conditions, while real-
world environments are dynamic.
• Scalability: Ensuring scalability while maintaining performance and reliability remains
a challenge.
• Data Consistency: Maintaining data consistency across distributed edge nodes is
complex.

III Real-World Applications:

285
International Journal of Research and Analysis in Science and Engineering

• Smart Cities: Real-time data processing from cameras and sensors for environmental
monitoring, public safety, and traffic control is made possible by edge computing. To
improve security and traffic flow, for instance, local data analysis is possible with traffic
signals and security cameras.
• Healthcare: By utilizing edge computing, medical sensors and wearable devices may
instantly analyze health data and issue alerts. Improved patient outcomes and prompt
treatments are made possible by local data processing and real-time vital sign
monitoring.
• Manufacturing: Edge computing supports predictive maintenance and real-time
monitoring of machinery and production lines. By analyzing data from industrial
sensors locally, manufacturers can detect anomalies, predict equipment failures, and
optimize production processes.
• Retail: Edge computing facilitates personalized customer experiences through real-
time data analysis and inventory management. Retailers can analyze customer behavior
locally to provide targeted promotions, manage inventory efficiently, and enhance the
overall shopping experience.
• Autonomous Vehicles: Edge computing allows self-driving cars to process sensor data
locally for faster decision-making. Real-time analysis of data from cameras, LiDAR,
and other sensors enables autonomous vehicles to navigate safely and respond to
changing road conditions.

IV Conclusion:

The reviewed studies highlight significant advancements in data management using edge
computing. Efficient data placement and retrieval, prediction-based data reduction, privacy-
preserving techniques, energy-efficient scheduling, and distributed deep learning are key
areas where innovative solutions are being developed. However, several gaps and
challenges remain, such as dynamic environment adaptation, scalability, data consistency,
prediction accuracy, resource constraints, data security, communication overhead, resource
allocation, energy consumption, and seamless integration of edge and cloud environments.
Addressing these challenges is crucial for optimizing the performance, security, and energy
efficiency of edge computing systems, paving the way for their widespread adoption in
various IoT and IIoT applications.

References:

1. Belcastro, L., Carretero, J., & Talia, D. (2024). Edge-Cloud Solutions for Big Data
Analysis and Distributed Machine Learning. Future Generation Computer Systems,
159, 323-326.
2. Frank, L. (2024). Edge AI: Deploying Models Directly on Edge Devices for Real-Time
Processing. Artificial Intelligence.
3. Nain, A., Sheikh, S., Shahid, M., & Malik, R. (2024). Resource Optimization in Edge
and SDN-Based Edge Computing: A Comprehensive Study. Cluster Computing.
4. Yang, L., Liao, Y., Cheng, X., Xia, M., & Xie, G. (2023). Efficient Edge Data
Management Framework for IIoT via Prediction-Based Data Reduction. IEEE
Transactions on Parallel and Distributed Systems, 34(12), 3309.

286
 Data Management in Edge Computing: Opportunities and Challenges

5. Meng, L., & Li, D. (2023). Novel Edge Computing-Based Privacy-Preserving

Approach for Smart Healthcare Systems in the Internet of Medical Things. Journal of
Medical Internet Research, 21(66).
6. Nair, A. K., Sahoo, J., & Raj, E. D. (2023). Privacy-Preserving Federated Learning
Framework for IoMT Based Big Data Analysis Using Edge Computing. Computer
Standards & Interfaces.
7. Liu, J., Yang, P., & Chen, C. (2023). Intelligent Energy-Efficient Scheduling with Ant
Colony Techniques for Heterogeneous Edge Computing. Journal of Parallel and
Distributed Computing, 172, 84-96.
8. Cui, T., Yang, R., Fang, C., & Yu, S. (2023). Deep Reinforcement Learning-Based
Resource Allocation for Content Distribution in IoT-Edge-Cloud Computing
Environments. Symmetry, 15(1), 217.
9. Khanh, Q. V., Nguyen, V.-H., Minh, Q. N., Van, A. D., Le, A. N., &Chehri, A. (2023).
An Efficient Edge Computing Management Mechanism for Sustainable Smart Cities.
Sustainable Computing: Informatics and Systems, 38, 100867.
10. Gómez Larrakoetxea, N., EskubiAstobiza, J., Pastor López, I., Sanz Urquijo, B., García
Barruetabeña, J., & Zubillaga Rego, A. (2023). Efficient Machine Learning on Edge
Computing Through Data Compression Techniques. IEEE Access,
10.1109/ACCESS.2023.3263391.
11. Li, M., Lei, H., Guo, H., & Shutaywi, M. (2023). Efficient Data Offloading Using
Markovian Decision on State Reward Action in Edge Computing. Journal of Grid
Computing, 21(2), 10.1007/s10723-023-09659-w.
12. Huang, L., Feng, X., Feng, A., & Qian, L. P. (2022). Distributed Deep Learning-Based
Offloading for Mobile Edge Computing Networks. Mobile Networks and Applications,
27(6).
13. Xie, J., Qian, C., Guo, D., Li, X., Shi, S., & Chen, H. (2019). Efficient Data Placement
and Retrieval Services in Edge Computing. IEEE International Conference on
Distributed Computing Systems (ICDCS), 106.

287
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

30. Development and Performance Testing of

Automatic Seed Sowing Agri Robot
Hamjadali A. Umachagi, Prashant Kadi,
Pavankumar Kulkarni
Assistant Professor, Department of EEE,
BLDEA’s V.P.Dr.P.G.H CET, Vijayapur.
Vinayak Sharma
B.E Student, Department of EEE,
BLDEA’s V.P.Dr.P.G.H CET, Vijayapur.
Saniya M. Patel
B. Tech CSBS Student, Center for UG/PG Studies,
Visvesvaraya Technological University, Belagavi.
Dr. Basanagouda F. Ronad
Associate Professor, Departmentof EEE,
Basaveshwar Engineering College (A), Bagalkot

ABSTRACT:

Agricultural development is one of the most powerful and vital sectors to end extreme
poverty. It is an allied sector and also the major livelihood provider in the country. In this
paper it is mainly focused on reducing the time taken for sowing the seeds and to minimize
the work of a farmer with minimum time along with the technology that is more easily
understood, implemented, and used by the farmers. Proposed model has a 4- 4-wheel robot
system and an Arduino Uno board which will control entire system process. If the seed box
is empty, then the ultrasonic sensor detects the level of the seed container and indicates its
status on the LCD display. The seed sowing machine is developed to get at an affordable
price. Also, the non-technical and unskilled farmer can also operate it very easily. The
single-row seeding mechanism is very simple to use and the various adjustments are made
with ease, which is maintenance-free. The system is powered with batteries, wheels are
provided for the rotation and a dc motor is inbuilt with those wheels to carry out the seed
sowing activity.

288
 Development and Performance Testing of Automatic Seed Sowing Agri Robot

KEYWORDS:

Agriculture, Agri-robot, Arduino Uno, Grafana, Influx DBLCD Display, IR Sensor, Sowing
machine, Ultrasonic Sensor.

I Introduction:

Agriculture contributes a significant figure level to our GDP. Sustainable changes are taking
place in agriculture techniques such as seeding, soil loosening, fertilizing, ploughing and
harvesting. Agriculture is crucial to our economic growth; hence it is very essential to
maximize agricultural productivity with good quality yields. Automating the process of day-
to-day seed plantation or seed sowing will minimize the farmer work and will be done
errorless. The small and compact-sized robot performs well and is of lightweight which
does not compact soil of the land [8].

A. Existing System:

The traditional way of showing the seed sowing is manual and faces plenty of the problems,
traditional techniques largely depend upon manpower and takes more time required with
more human efforts to accomplish the work. Humans cannot work for long hours so need
to rest and also cannot work in environments which is hazardous. To complete one particular
work in the agriculture sector, we need to have better manpower which is accomplished by
an automated robot which does soil loosening and seed sowing task. The traditional way
suffers from plenty of problems. So, the main aim of the proposed work is to minimize the
human effort as well as the time requirement and to maximize the crop production with fine
accuracy [2].

B. Proposed System:

The proposed system consists of a system that provides fast soil loosening, digging, seeding,
and closing [2]. The robot is controlled by ATMEGA328PMicrocontroller with the help of
a command received by the HC-05 Bluetooth Module robot which moves forward,
backward, left, and right. The 12V adapter is used to energize the system. The Arduino Uno
board is used to operate the complete system [9]. The three DC motors are used in robot
applications whereas two motors are used as the wheel of the robot for the rotation of the
robot and one DC motor for the seeding process where we can use the oneL293D motor
driver motor to control two motors and another L293D motor driver is used to control the
speed of seeding process. With the help of the Bluetooth module, agri-robot can operate as
per user requirement to turn left, right, forward, and backward. The saw tooth mechanism
set up is used to loosen the land soil and digging, and a mechanical setup are done for closing
the soil.

II System Architecture:

The overall objective of the proposed work is represented in the block diagram which
depicts how all different components are interconnected to each other to perform their own

289
International Journal of Research and Analysis in Science and Engineering

role to finish the particular work of this proposed work. Using a 12V adapter, the 12V supply
is provided to the motor drive and for the Arduino Uno board a 9V battery is provided.

The code is written using 'Embedded C' language in the Arduino IDE Software which is
given to Arduino as the input using a USB Arduino cable after that the Arduino Uno board
is used to interface with all connected components such as L293D Motor Drive, Bluetooth
module, Ultrasonic Sensor along with the display. An ultrasonic sensor is used to detect the
availability of seed present in the container and the data will be displayed on the display
screen.

A. Block Diagram:

Figure 2.1: Block Diagram of the Proposed System.

Figure 2.1 represents the block diagram of the proposed model which consists of motors for
different agriculture purpose, various sensors, Node MCU, Bluetooth module and LCD
display.

B. Proposed Working Model:

290
 Development and Performance Testing of Automatic Seed Sowing Agri Robot

Figure 2.2: Robot Seed Boxis Empty

Figure 2.2 illustrates the working model with empty seed status which indicates to add the
seeds in the box for sowing.

Figure 2.3: Robot Box has Seed

Figure 2.3 illustrates the working model with the status of loaded seed boxusing ultrasonic
sensor for further sowing.

III Technology Used:

A. ESP8266 Module:

The ESP8266 module is a multifaceted and cost-effective microcontroller incorporating

built-in Wi-Fi capabilities. Equipped with a TCP/IP stack, this module seamlessly connects
to Wi-Fi networks and facilitates communication with other devices over the internet. Our
utilization of the ESP8266 encompasses a wide range of functionalities, including sensor
data collection, motor control, and bidirectional data exchange with databases and cloud
services.

B. Arduino IDE Software:

The Arduino IDE (Integrated Development Environment) software incorporates essential

components such as a compiler, a source code editor for scripting, a debugger, and a builder.
This software is utilized for uploading code to any Arduino family board. It is also employed
to establish a connection between the Arduino Uno board and a laptop or computer using a
USB cable to upload a written program and to facilitate communication with it.

C. Ultrasonic Sensor:

The ultrasonic sensor is designed to measure distance and is utilized in this proposed project
291
International Journal of Research and Analysis in Science and Engineering

to detect the presence of seeds in the seed container. The sensor comprises two components:
the TRIG and ECHO modules.

The TRIG module emits an ultrasound signal at a frequency of 40kHz, which then
propagates through the air medium. When an object or obstacle obstructs its path, the signal
reflects back to the module. The pin configuration of the Ultrasonic HC-SR04 features VCC,
GND, TRIG, and ECHO pins. The VCC pin is used to supply the necessary voltage, and the
TRIG and ECHO pins can be connected to any available Digital I/O pin on the Arduino Uno
board.

1. Grafana

Grafana is utilized as a central component of agri-robot model for monitoring and

visualization process. Specifically, Grafana enables to aggregate and display real-time and
historical data from the robot, including performance metrics and sensor readings. This
allows to gain valuable insights and identify trends, ultimately aiding in informed decision-
making and performance optimization. Additionally, the customizable nature of Grafana
allows to tailor the visualizations to agri-robot requirements, ensuring a clear and
meaningful representation of the data [11].

2. Influx DB:

Influx DB is a specialized time-series database designed to efficiently store and retrieve data
that changes over time. It is specifically optimized for managing data points with
timestamps, making it well-suited for system's requirements in tracking changes and trends
over time. Influx DB is commonly utilized in monitoring and analytics scenarios, such as
tracking sensor data, monitoring system performance metrics, or logging events in real time
[12,13].

3. I2C LCDD is play:

The 16X2 LCD Display is connected to an I2C Module to transform it into a 16X2 I2C
LCD Display. A notable distinction between the two displays is the reduced wiring
requirement of the I2C Display, which necessitates only four wires: VCC, GND, SCL, and
SDA.

This display system is designed to indicate the presence of seeds within a container using
an ultrasonic sensor and an Arduino Uno board. When seeds are detected, the display will
show "Box Has Seed" and "Happy Sowing," whereas in the absence of seeds, it will display
"Seed Box is Empty" and "Add Seeds in Box."

4. L293D Motor Driver:

In the proposed application, two motor drivers are used to regulate the speed and direction
of the robot, as well as to control the seeding process. These motor drivers are operated
using a +12V power supply and are equipped with the widely used L293 IC for seamless
control of the connected motors.

292
 Development and Performance Testing of Automatic Seed Sowing Agri Robot

Figure 3.1: Illustration of Motor Direction change

The direction of the DC Motor can be controlled by the polarity of its input terminal or
voltage. This technique is done using H-Bridge.

IV Flowchart of the Proposed System:

Figure 4.1: Flow Chart

293
International Journal of Research and Analysis in Science and Engineering

The flow of operations for the robot begins with its initialization. Initially it establishes a
connection to a Wi-Fi network to enable communication capabilities. Followed by this, it
configures the Influx DB, a vital component which facilitates the linkage between robot and
the cloud-based database. If the specified database does not already exist, the system
proceeds to create it.

Subsequently, Grafana, a data visualization tool, establishes connectivity with the database.
This enables the seamless display of all data collected by the robot in a user-friendly format.

The robot functions in two distinct modes: normal mode and seeding mode. In the normal
mode, the robot operates analogous to a conventional vehicle without engaging in any
seeding activities. Conversely, in seeding mode, the robot's cultivators descend to the
ground, creating furrows at predefined intervals for sowing seeds. After completing this
task, robot closes the dug portions, thereby completing the seeding process.

V Output:

Figure 5.1 and Figure 5.2 represents the output of the agri-robot. The spacing and depth
between seed to seed varies from crop to crop.

Table 5.1: The distance between the two different seeds

Seeds Distance
Soyabean 18cm
Groundnuts 15cm
Jawar 12cm

Figure 5.1: Robot Seeding Process

294
 Development and Performance Testing of Automatic Seed Sowing Agri Robot

Figure 5.2: Robot Seeding Process Back View

VI Results:

Influx DB is used to store the entire data.

Figure 6.1: Influx DB Data Explorer

Figure 6.1 represents Influx DB window which is a powerful time-series database which
is used to manage and store all the required data. Its optimized scheme and efficient
design make it ideal for handling huge volumes of time-stamped information. With Influx
DB, easy storage and retrieval of the data points can be done, it enables to visualize and
analyze trends and time-based patterns effectively. Its adaptable query language and data
model provides the freedom to arrange and retrieve data in accordance with user unique
requirements. Influx DB's powerful features it has enabled to create scalable and reliable
data storage system [12,13].

295
International Journal of Research and Analysis in Science and Engineering

Figure 6.2: Grafana Dashboard

The illustration in Figure 6.2 depicts the seeding mode within Grafana, an advanced
platform for data visualization and monitoring. Grafana is seamlessly integrated with
Influx DB to deliver insightful visual representations of stored data. Through the
combined capabilities of Grafana and InfluxDB, it has enabled to create dynamic,
interactive dashboards that present both historical and real-time data. Grafana offers a
diverse array of visualization options, including tables, graphs, and charts, enabling us to
present data in a visually engaging and easily comprehensible manner [11].

Figure 6.3: Output representation of all the connected sensors on the serial monitor
of the Arduino IDE.

296
 Development and Performance Testing of Automatic Seed Sowing Agri Robot

VII Conclusion:

An automatic seed sowing robot has been successfully developed to streamline various
agricultural tasks and enhance overall efficiency. From soil preparation to seeding and
monitoring, this technology aims to minimize human effort, reduce time constraints, and
ultimately maximize crop production. The incorporation of advanced performance monitoring
technology, such as Grafana for visualization, ensures comprehensive oversight and control.
This Agri-robot marks a significant step forward in agricultural automation and the pursuit of
sustainable, high-yield farming practices.

VIII Future Scope:

This proposed work can be upgraded with several seeding arms, which can be extended up to
4 to 6 rows in single time. This can furthermore reduce the time for the production of crops.

References:

1. Hamjad Ali Umachagi, Pavankumar Kulkarni, and Muttanna Bilagikar “Implementation

of Automated Water Supply and Distribution using PLC and SCADA”, 2020 IEEE
Bangalore Humanitarian Technology Conference (B-HTC), 08-10 October 2020.
2. M. Aravind Kumar, Akkarapalli Sanjeev Reddy, andK. Sagadevan"
AUTOMATICSEED SOWING&IRRIGATION AGRIBOT USING ARDUINO”,
International Journal of Pure and Applied Mathematics, vol. 119, no. 14, 2018.
3. Sidhanth Kamath, Kiran K Kharvi, Abhir Bhandary, and Jason Elroy Martis, "IoT based
Smart Agriculture", International Journal of Science Engineering and Technology
Research (IJSETR), vol. 8, no. 4, April 2019.
4. S Praseenal, S Sanjana, S M Thejaswini, and M Senthamil Selvi, "Sensor-Based
AGROBOT for Sowing Seeds", International Research Journal of Engineering and
Technology (IRJET), vol. 06, no. 03, Mar 2019.
5. S. Umarkar and A. Karwankar, "Automated seed sowing agribot using Arduino", 2016
International Conference on Communication and Signal Processing (ICCSP), pp. 1379-
1383, 2016.
6. G. Sushanth and S. Sujatha, "IOT Based Smart Agriculture System," 2018 International
Conference on Wireless Communications, Signal Processing and Networking
(WiSPNET), Chennai, India, 2018, pp. 1-4, doi: 10.1109/WiSPNET.2018.8538702.
7. D. S. Babu, V. G. Bhavani, S. Jain, S. N. Baliga, S. Srikanth and S. Gurugopinath,
"Krrushikar: Design and Development of a Seed Sowing Planter Bot and a Smart
Greenhouse," 2021 IEEE International Conference on Electronics, Computing and
Communication Technologies (CONNECT), Bangalore, India, 2021, pp. 1-6, doi:
10.1109/CONECCT52877.2021.9622733.
8. K. Ramesh, K. T. Prajwal, C. Roopini, M. Gowda M.H. and V. V. S. N. S. Gupta, "Design
and Development of an Agri-bot for Automatic Seeding and Watering Applications,"
2020 2nd International Conference on Innovative Mechanisms for Industry Applications
(ICIMIA), Bangalore, India, 2020, pp. 686-691,
doi: 10.1109/ICIMIA48430.2020.9074856.
9. G. Sushanth and S. Sujatha, "IOT Based Smart Agriculture System," 2018 International
Conference on Wireless Communications, Signal Processing and Networking
(WiSPNET), Chennai, India, 2018, pp. 1-4, doi: 10.1109/WiSPNET.2018.8538702.
297
International Journal of Research and Analysis in Science and Engineering

10. B. Ragavi, L. Pavithra, P. Sandhiyadevi, G. K. Mohanapriya and S. Harikirubha, "Smart

Agriculture with AI Sensor by Using Agrobot," 2020 Fourth International Conference on
Computing Methodologies and Communication (ICCMC), Erode, India, 2020, pp. 1-4,
doi: 10.1109/ICCMC48092.2020.ICCMC-00078.
11. Venkatramulu, s & Phridviraj, M.S.B. & Srinivas, C. & Rao, Vadithala. (2021).
Implementation of Grafana as open-source visualization and query processing platform
for data scientists and researchers.
12. Giacobbe, Maurizio & Di Pietro, Riccardo & Longo Minnolo, Antonino & Todaro,
Marco & Puliafito, Antonio. (2018). Using Influx DB Time Series Database to
Manage Smart Environments.
13. Lin, Longhan. (2024). Storage and Data Processing Technology in the Era of
Information Explosion: from Traditional to Edge Computing.
14. Giacobbe, Maurizio & Di Pietro, Riccardo & Longo Minnolo, Antonino & Todaro,
Marco & Puliafito, Antonio. (2018). Using InfluxDB Time Series Database to
Manage Smart Environments.
15. Avula Likitha, B. Mamatha, Agamanthi Sai kiran, Dondeti Pranitha,” IoT Based Smart
Agriculture and Automatic Seed Sowing Robot”, International Journal of Resource
Management and Technology, ISSN No:0745-6999.

298
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

31. Edge Computing Based Smart Health Care

System
Sadashiv Badiger
GH Raisoni College of Engineering and Management,
Pune, Maharashtra, India.
Chayalakshmi C. L., Mahabaleshwar S. K.,
Rajani S. Pujar
Basaveshwar Engineering College,
Bagalkote, Karnataka, India.

ABSTRACT:

In the realm of emergency medical services, timely and accurate health data is crucial for
ensuring the well-being of patients in transit to healthcare facilities. The proposed system
integrates edge computing technologies to enable real-time data processing and analysis,
facilitating swift decision-making and improved patient care. Our smart ambulance is
equipped with various IoT sensors and medical devices that continuously monitor the
patient's vital signs, transmitting critical data to health care providers instantly. This setup
allows for early diagnosis and intervention, significantly improving patient outcomes.
Additionally, the edge computing frame work ensures minimal latency in data transmission
and processing, even in remote or bandwidth-limited areas. The proposed system's
architecture, implementation details, and performance evaluation are thoroughly
discussed, demonstrating its potential to revolutionize emergency medical services. Our
findings suggest that Edge Computing-based Smart Ambulances can play a crucial role in
modernizing emergency health care, offering a promising solution to the challenges faced
in urgent medical scenarios.

KEYWORDS:

Smart Ambulance, Edge Computing, IoT Sensors.

I Introduction:

Medical emergencies demand rapid and efficient response systems to save lives and reduce
the severity of injuries. Traditional ambulance services, while vital, often face challenges
such as delays in diagnosis, limited real-time data access, and communication issues
between the ambulance and health care facilities [1].

299
International Journal of Research and Analysis in Science and Engineering

These limitation scan critically impact the quality of care provided to patient’ senrouted to
hospitals. Recent advancements in technology offer promising solutions to these challenges.

Figure 1: Edge Computing Architecture for Smart Ambulance

The figure 1 illustrates the architecture of the Edge Computing-based Smart Ambulance
system, comprising three distinct layers: The Device Layer, the Edge Layer, and the Cloud
Layer. At the Device Layer, ambulances are equipped with sensors and controllers that
continuously monitor and collect patient data. This data is transmitted to the Edge Layer,
where edge nodes or servers handle data processing and control response in real-time,
ensuring low latency and immediate decision-making. The Edge Layer facilitates efficient
communication between the ambulances and the cloud. The Cloud Layer, consisting of
cloud servers and data centers, performs extensive data computing and big data processing,
enabling advanced analytics and long-term data storage [2]. This layered approach ensures
a robust, scalable, and responsive system capable of delivering timely medical interventions
during emergencies. The proposed methodology delves into an innovative ambulance
system tailored for swift response to emergencies, especially in scenarios with multiple
injured patients like major accidents or natural disasters. It integrates cutting-edge edge
computing technology, including machine learning algorithms and Raspberry Pi devices, to
enhance speed and accuracy in collecting and analyzing patient data for prompt treatment
suggestions. Cloud connectivity facilitates efficient resource management and long-term
hospital planning. A prioritization mechanism ensures urgent cases receive immediate
attention during transportation, supported by specialized training for ambulance staff to
effectively handle high-stress situations.

II Literature Survey:

Integrating edge computing with smart ambulance systems offers significant benefits for
patientcare and outcomes. By leveraging technologies like IoT and real-time data analytics
[4], these systems enable the collection and transmission of vital patient data to healthcare
providers, allowing for better preparedness and reduced treatment delays [4].

300
 Edge Computing Based Smart Health Care System

Additionally, the utilization of Multi-Access Edge Computing (MEC) enhances the

efficiency of health care services by providing extremely low latency, substantial
bandwidth, and optimized resource usage [5]. This integration facilitates continuous
monitoring of emergency patients, addressing the lack of real-time risk assessment and
dynamically providing necessary medical interventions [6]. Overall, the combination of
edge computing with smart ambulance systems improves response times, enhances patient
care, and ultimately contributes to better health outcomes in emergency situations.

Edge Computing Based Smart Ambulance systems can significantly optimize medical
emergency response times in urban areas by integrating various technologies. These
systems utilize RFID, GPS, LTE, IoT, and real-time data analytics to stream line ambulance
transit and enhance emergency services [7]. By leveraging multi-sensor integration,
including RFID sensors, cameras, and microphones, these smart ambulance scan
communicate with traffic signals, detect congestion levels, and identify ambulance presence
in traffic, enabling dynamic rerouting to less congested routes and triggering traffic signal
adjustments for prioritized passage [8].

Additionally, IoT platforms like Blynk facilitate real-time tracking of ambulances, ensuring
swift identification of the nearest hospital and enabling vital data transmission to healthcare
providers before the patient's arrival [9]. Furthermore, the incorporation of IoT devices
within ambulances allows for the collection and transmission of vital patient signs,
preparing medical personnel in advance and reducing delays in treatment [10]. By
combining these technologies, Edge Computing Based Smart Ambulance systems
revolutionize emergency response, ensuring prompt and efficient medical care delivery in
urban settings. Edge Computing Based Smart Ambulance services, as proposed in various
research papers, significantly enhance patient outcomes in remote areas by leveraging IoT
technologies for real-time monitoring and data transmission [11].

These services integrate wear able devices and IoT nodes to continuously monitor patients
during transit, enabling immediate medical interventions based on real-time risk
assessments [12]. Additionally, the incorporation of GPS systems in ambulances allows for
swift identification of the nearest hospitals, ensuring timely access to necessary medical
care [13]. The utilization of cutting-edge software features and Lab-On-Chip technology
within Electric Ambulances further optimizes response times and facilitates on-the-spot
health examinations, ultimately improving treatment efficacy and patient outcomes in
emergency situations [14]. By enhancing communication between emergency responders
and health care facilities through information technology, these smart ambulance systems
streamline the decision-making process and improve overall rescue efficiency, especially in
remote or rural areas [15]

III Methodology:

Following an extensive examination of various research papers, it has become clear that
emergency response teams encounter numerous challenges during major accidents or
natural disasters. One of the primary obstacles is the large number of injured individuals,
which often exceeds the capacity of existing ambulance systems, resulting insignificant
delays in providing urgent medical assistance.

301
International Journal of Research and Analysis in Science and Engineering

Figure 2: Block Diagram of Proposed Methodology

A. Workflow:

Figure 2 illustrates the work flow and components of the Edge Computing-based Smart
Ambulance system.

Central to the system are various sensors with in the ambulance, including temperature
sensors, heartrate sensors, pulse oximeter sensors, and EMG sensors, all of which monitor
the patient's vital signs in real-time. The data collected by these sensors is processed by an
ATM ega 328 micro controller, which compiles the information into a CSV file format for
efficient handling and transmission. This patient vital health data is then sent to a Raspberry
Pi4, which uses machine learning algorithms trained on a dataset of patient health records
to analyze the data and propose potential treatments. The processed in formation and
proposed treatments are forwarded to a cloud data center, enabling seamless communication
and data storage. Simultaneously, the data is transmitted to the hospital, ensuring that
medical personnel are well-informed about the patient's condition before arrival. This
integration of real-time monitoring, edge computing, and machine learning with in the smart
ambulance system enhances the ability to provide timely and accurate medical care during
emergencies.

B. Sensors:

1. Temperature Sensor: The DS18B20 digital temperature sensor is renowned for its high
accuracy and versatility. It operates within a broad temperature range of -55° C to +
125° C, with an impressive accuracy of ± 0.5°C from-10°C to +85°C. The sensor offers
programmable resolution from 9 to12 bits, allowing for flexibility based on the
application's precision requirements. It communicates via a1-Wire protocol, requiring
only one data line for communication and power, and supports a supply voltage range
of 3.0V to 5.5V.

302
 Edge Computing Based Smart Health Care System

2. Heart Rate Sensor: The Pulse Sensor, or heart rate sensor, is a highly efficient and
accurate device designed to measure pulse rates using the photo plethysmography
(PPG) method. Operating at a voltage range of 3.3V to 5V, it consumes less than 4m A
of current, making it energy-efficient. It provides an analog output signal suitable for
direct interface with microcontrollers. The sensor can measure heart rates ranging from
0 to 220 beats per minute (BPM) with an accuracy of ±2 BPM and a response time of
less than100ms.
3. Pulse Oximeter Sensor: The MAX 30100 Pulse Oximeter Sensor is a highly integrated
module designed for accurate measurement of oxygen saturation (SpO2) and heartrate.
It operates at 1.8V for the core and 3.3V for I/O, with atypical current consumption of
600µ A in normal mode and just 0.7µA in shutdown mode, making it ideal for battery-
powered applications. The sensor utilizes a reflective photoplethysmography (PPG)
method with red (660 nm) and infrared (880 nm) LEDs, providing an SpO2
measurement range of 0% to 100% with an accuracy of ± 2% in the 70% to 100% range,
and a heart rate range of 30 to 240 BPM with ± 1 BPM accuracy.
4. EMG Sensor: The electromyography (EMG) sensor's technical specifications include
a frequency response range typically from 10 Hz to 1 kHz and a signal amplitude range
of ± 1 mV to ±10 mV, ensuring accurate detection of muscle electrical activity. It
features a high input impedance of over 10 MΩ and a dynamic range of 20-40 dB, which
supports a signal-to-noise ratio of 60-80 dB for clear data acquisition. The sensor
operates with a low power supply of 3-5 V DC and offers connectivity through analog
or digital interfaces like USB or Bluetooth.

C. Microcontroller ATMEGA328:

The AT mega 328 microcontrollers also called Arduino, a popular member of the Atmel
AVR family, is renowned for its efficiency and versatility in embedded systems. It features
an 8-bit AVR RISC architecture with a clock speed of up to20MHz, providing a robust
processing capability for various applications. The microcontroller includes 32 KB of flash
memory for program storage, 1 KB of SRAM for dynamic data, and 2KB of EEPROM for
non-volatile data retention. It offers 23I/Opins, which can be used for digital input/output,
along with six analog-to-digital converter (ADC) channels with a10-bitresolution. The AT
mega 328 also supports serial communication via USART, SPI, and I2C interfaces, and
operates on a voltage range of 1.8V to 5.5V. Its low power consumption and broad range of
features make it a popular choice for hobbyists and professionals alike, especially in
Arduino-based projects.

D. Implementing ML Algorithm on Raspberry Pi 4:

The Raspberry Pi 4 Model B is a powerful single-board computer equipped with aqua core
ARM Cortex – A 72 processor running at 1.5GHz, and it offers 2GB, 4GB, or 8GB of
LPDDR4- 3200 SDRAM, depending on the model. It features dual HDMI ports for 4K video
output, multiple USB 3.0 and USB 2.0 ports, and Gigabit Ethernet for high-speed network
connectivity. Implementing machine learning (ML) algorithms on the Raspberry Pi 4 is
feasible thanks to its enhanced processing power and RAM, allowing for the execution of
light weight models and inference tasks.

303
International Journal of Research and Analysis in Science and Engineering

In our proposed methodology we have leveraged popular libraries such as Tens or Flow Lite
and PyTorch, which are optimized for edge devices, to deploy pre-trained models or perform
on-device training. The Raspberry Pi 4’s GPIO pins and extensive community support
further enable integration with sensors and peripherals, making it a versatile platform for
ML applications ranging from image recognition to predictive analytics in resource-
constrained environments.

E. Trained Data Set and Patient Vital Data:

The trained data set refers to a collection of historical patient health records used to develop
machine learning models. This data set typically includes, Demo graphic Information, Vital
Signs, Medical Conditions, and Sensor Data. The purpose of this trained data set is to provide
a comprehensive foundation for training machine learning algorithms. These algorithms
learn patterns and correlations between the input features and the output labels. Once trained,
the machine learning models can predict patient conditions and suggest appropriate
treatments based on new patient data.

The patient vital health data captured in a CSV file includes real-time health metrics recorded
by various sensors attached to the patient. The CSV file contains structured data such as,
Time stamp, Temperature Readings, Heart Rate Readings, Oxygen Saturation Levels and
EMG Readings.

This CSV file serves as the primary input for the machine learning models running on the
Raspberry Pi 4. The data is processed and analyzed in real-time to detect anomalies, predict
potential health issues, and propose personalized treatments. The system ensures that critical
health data is continuously monitored, allowing for timely interventions and improved
patient outcomes. Additionally, the data is uploaded to a cloud data center, enabling remote
access and further analysis by healthcare professionals in a hospital setting.

F. Cloud Data Center:

The cloud data center serves as a centralized and secure repository for storing vast amounts
of patient health data, enabling remote access and real-time monitoring by healthcare
professionals. It provides scalable storage solutions, advanced data analytics capabilities,
and supports the training and updating of machine learning models to ensure accurate and
up-to-date health predictions. Additionally, the cloud data center facilitates seamless
integration with hospital systems, robust security protocols, and compliance with healthcare
regulations, enhancing the overall functionality and security of the healthcare system.

IV Result and Analysis:

In this section, we present the outcomes of our opposed system for monitoring patient vitals
and suggesting treatment using machine learning algorithms. The system comprises multiple
sensors, a microcontroller, and a Raspberry Pi for data processing, as illustrated in the figure
2.

304
 Edge Computing Based Smart Health Care System

A. Sensors Output:

The system employs various sensors (temperature, heartrate, pulse oximeter, and EMG) to
collect patient vital signs. Data from these sensors is transmitted to the ATMEGA 328
microcontroller, which aggregates the readings and formats them into a CSV file for further
processing. The following figure 3 shows the interfacing of the sensor with microcontroller.

Figure 3: Interfacing All the Sensors to The Microcontroller

The following are steps to access sensor data from an Arduino and read it using a Raspberry
Pi.

a) Arduino Side (Pseudocode):

Step1: Setup Serial Communication, initialize serial communication at 9600 baud rate and
continuously read data from temperature, heart rate, pulse oximeter, and EMG sensors.

Step 2: Send data over serial, format the sensor data into a read able string and send the
formatted string over the serial port.

Step 3: Send Data over Serial: Format the sensor data into are a dabble string. Send the for
matted string over the serial port.

b) Raspberry Pi Side (Pseudocode):

Step1: First Setup Serial Connection, import necessary libraries and establish a serial
connection to the Arduino.

Step 2: Read data from serial continuously check if data is available on the serial port. Read
the incoming data line by line and Print or process the received data.

Step3: Store Data, open a CSV file for writing and Write the received data into the CSV
file.

305
International Journal of Research and Analysis in Science and Engineering

Figure 4: Sensor data from Raspberry Pi serial port

B. Predictive Analysis:

The dataset to be analyzed comprises various patient health parameters, including

temperature, heart rate, oxygen level, and muscle strength. These parameters will be utilized
for analysis and exploration in the study. The dataset to be trained is depicted below in table
1.

Table I: Training data Set

Following is the evaluation of different machine learning algorithms including SVM

(Support Vector Machine), Decision Tree, Logistic Regression, and Naïve Bayes, it was
concluded that the Decision Tree algorithm exhibited the highest accuracy for the health
dataset. The table 2 presents the different parameters used for calculating accuracy.

306
 Edge Computing Based Smart Health Care System

Table II: Performance Comparison of Various Machine Learning Algorithms Based

On Accuracy Metrics

C. Running ML Applications on Raspberry Pi:

Running machine learning (ML) applications on a Raspberry Pi for the application shown
in the figure 21. 2 involves several steps, including setting up the Raspberry Pi, collecting
and preprocessing data, training a machine learning model, and deploying the model to
make predictions. Following steps illustrate the running machine learning applications are
Raspberry Pi.

Step1: Set Up Raspberry Pi - Ensure Raspberry Pi is running the latest version of the
Raspbian OS. Update the system packages to the latest versions. Install Python and
necessary libraries such as Num Py, pandas, scikit-learn, and others required form a chine
learning tasks.

Step2: Collect and Prepare data -Interface all the sensors to Arduino board, using the
USB cable connect the Arduino board to the Raspberry Pi. Write a script on the raspberry
to read the serial data from the Arduinos and save it into CSV file. Load the collected data,
clean it, and prepare it for training. This might involve normalizing or scaling the data,
handling missing values, and splitting it into training and testing sets.

Step3: Train machine learning model - From the table 2, it is analyzed that decision tree
algorithm performs very well for such application. After selecting appropriate machine
learning algorithm, train the model using training data. Test the model on the testing set to
evaluate its performance. Save the trained model for further preprocessing steps.

Step4: Deploy and Run ML model on Raspberry Pi -We have written python script to
load the saved model on the raspberry pi. Continuously read the data from the sensors
connected to the Arduino in real time. Trained models are used to make the prediction based
on the real-time data. Display the proposed action in the ambulance.

307
International Journal of Research and Analysis in Science and Engineering

Step5: Send data to cloud and hospital - Send data to the cloud data center for storage
and further analysis. Also ensure that the data stored in the cloud is accessible to hospital
system for monitoring and further analysis by medical professionals.

D. Cloud Connectivity:

First, vital health data collected from sensors and wear able devices is processed at the edge
(on the ambulance) to reduce latency. The processed data is then securely transmitted to Ad
fruit IO, a cloud-based IoT platform, where it is visualized on a real-time dashboard. This
dashboard provides medical personnel and emergency responders with immediate access to
critical health information. The dashboard's user-friendly interface and real- time updates
enable a more effective response, ensuring that patients receive timely and appropriate care
during emergencies. This integration of edge computing and Ad a fruit IO dashboard
enhances the efficiency of ambulance services and ultimately improves patient outcomes.
Figure 5 shows the

Figure 5: The above figure shows the IoT dashboard of Ad a fruit Cloud and it shows
the streamed data of (a) Pulse oximeter sensor (b) Heart rate sensor (c) Temperature
Sensor and (4) EMG Sensor

308
 Edge Computing Based Smart Health Care System

Future Scope:

Cloud capabilities provide centralized storage and extensive data analysis, supporting long-
term trends and large-scale analytics. Edge computing, meanwhile, processes data locally,
reducing latency and enabling real-time decisions, especially in time-sensitive situations.
By balancing immediate, localized processing with large-scale cloud analysis, these
technologies complement each other, offering both instant responses and comprehensive
data insights. Enhancing data security through encryption and exploring blockchain will
protect patient information, while user-friendly interfaces will improve system accessibility.
Adopting interoperability standards will ensure continuous improvement and adaptability
to various medical conditions, ultimately transforming patient monitoring and healthcare
outcomes.

References:

1. R. T. Iype, "Autonomous Ambulance Management System with Real-Time Patient

Monitoring using IoT," 2019 IEEE 5th International Conference for Convergence in
Technology (I2CT), Bombay, India,2019, pp.1-4.
2. J. Leeand W. Na, "A Surveyon Vehicular Edge Computing Architectures," 2022 13 th
International Conference on Information and Communication Technology
Convergence (ICTC), Jeju Island, Korea, Republic of, 2022, pp.2198-2200.
3. Jouini, O.; Sethom, K.; Namoun, A.; Aljohani, N.; Alanazi, M.H.; Alanazi, M.N, “A
Survey of Machine Learning in Edge Computing: Techniques, Frameworks,
Applications, Issues, and Research Directions,” Technologies 2024,12,81.
4. Shubham & Rajiwade, Vikram, “The Smart AMBULANCE Service”, International
Journal of Scientific Research in Science, Engineering and Technology2020, pp.70-73.
5. Xavier, R.; Silva, R. S.; Ribeiro, M.; Moreira, W.; Freitas, L.; Oliveira-Jr, A. Integrating
Multi- Access Edge Computing (MEC) into Open 5G Core. Telecom 2024,5,433-450.
6. Mukhopadhyay, A.; Remanidevi Devidas, A.; Rangan, V. P.; Ramesh, M.V.A QoS-
Aware IoT Edge Network for Mobile Telemedicine Enabling In-Transit Monitoring of
Emergency Patients. Future Internet2024,16,52.
7. Krishnan, Suren & Thangaveloo, Rajan & Sindiramutty, Siva Raja, “Smart Ambulance
Traffic Control System. Trends in Under Graduate Research”, 2021, 4.10.33736
/tur.2831.2021.
8. Deepika Sarpal, Yatharth Asthana and Dr. Malaya Kumar Hota, Reviewon “Smart
Traffic Management System for Ambulance”, INTERNATIONAL JOURNAL OF
ELECTRICAL ENGINEERING AND TECHNOLOGY. 11.10.34218/IJEET.11.6.20
20.001.
9. Azmin, A. & Abdullah, Samihah & Faiza, Zafirah & Ahmad Ahmad Fauzi, Najwa
Rawaida & Shahanim, Nor & Rahiman, Wan, “Fingerprint Sensor Integration in Smart
Healthcare Emergency App: Enhancing Ambulance Navigation through Emergency
Route Highlighting”85-89.85-89.10.1109/CSPA60979.2024.10525512.
10. C. Thaijiam,"A Smart Ambulance with Information System and Decision-Making
Process for Enhancing Rescue Efficiency," in IEEE Internet of Things Journal, vol. 10,
no. 8, pp. 7293-7302, 15 April15,2023, doi:10.1109/JIOT.2022.3228779.
11. Mukhopadhyay, A.; Remanidevi Devidas, A.; Rangan, V. P.; Ramesh, M.V. “A QoS-
Aware IoT Edge Network for Mobile Telemedicine Enabling In-Transit Monitoring of
Emergency Patients”, Future Internet 2024,16,52.
309
International Journal of Research and Analysis in Science and Engineering

12. Nallusamy, Selvakumar, “Transforming Emergency Medical Services with Electric

Ambulance Technology”, International Journal for Research in Applied Science and
Engineering Technology, 2023 11.10.22214/ijraset.2023.57201
13. C. Thaijiam,"A Smart Ambulance with Information System and Decision-Making
Process for Enhancing Rescue Efficiency," in IEEE Internet of Things Journal, vol. 10,
no. 8, pp. 7293-7302, 15 April15,2023
14. E. Kyriacou, R. Constantinou, C. Kronis, G. Hadjichristofi and C. Pattichis,
"eEmergency System to Support Emergency Call Evaluation and Ambulance Dispatch
Procedures," 2020 IEEE 20th Mediterranean Electro Technical Conference
(MELECON), Palermo, Italy, 2020, pp.354-357
15. K. Devibalan, S. Brindha, R. Sreeraman, M. E. Thrinam Vishwakumaar and J.
Muralidharan, "Smart Ambulance System using IoT," 2024 International Conference
on Communication, Computing and Internet of Things (IC3IoT), Chennai, India,2024,
pp.1-6.

310
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

32. Integrating Block chain in EV Charging

Systems for Secure and Efficient Infrastructure
Preetam Kanal, Akshata Dhagade, Bhagyashree Belagali,
Darini Budihal, Rajani S. Pujar
Electronics and Communication Engineering Department,
Basaveshwar Engineering College,
Bagalkote, Karnataka, India

ABSTRACT:

With the growing adoption of electric vehicles (EVs) and the pressing need for secure and
efficient charging infrastructures, ensuring the security of communication protocols within
EV charging management systems is crucial. This study applies Burrows–Abadi–Needham
(BAN) Logic to analyse the security of a proposed protocol for EV charging management.
The protocol facilitates mutual authentication between the EV, the charging station, and
the backend management system, ensures that only authorized EVs can initiate charging
sessions, and maintains the integrity and confidentiality of all exchanged messages. The
BAN Logic analysis confirms the robustness and security of the protocol, highlighting its
effectiveness in preventing common security threats such as replay attacks, impersonation,
and unauthorized access. This formal verification underscores the importance of BAN Logic
in enhancing the reliability and trustworthiness of EV charging infrastructures.

KEYWORDS:

Index Terms—Electric Vehicles (EVs); EV Charging Management System; Burrows–

Abadi–Needham Logic (BAN)

I Introduction:

As electric vehicles (EVs) and the Internet of Things (IoT) continue to proliferate, their
integration into smart grids offers a transformative approach to managing distributed energy
and electricity generation [1]. This integration is manifested through various vehicular
systems, including vehicular ad hoc networks (VANETs), vehicle-to-grid (V2G) systems,
vehicle-to-vehicle (V2V) communications, and the Internet of Vehicles (IOV). These
systems rely on a diverse array of communication and measurement sensors—such as speed
detectors, GPS modules, Bluetooth, Wi-Fi, and on-board units (OBUs)—to gather and
transmit critical data related to vehicle speed, location, identity, and movement.

311
International Journal of Research and Analysis in Science and Engineering

However, the transmission of such sensitive data over public networks introduces
significant security risks. The data is susceptible to interception, modification, and misuse
by malicious actors. To mitigate these risks and ensure secure communication, robust
mutual authentication and key agreement mechanisms are crucial.

Over the past decades, numerous authentication and key agreement schemes have been
proposed to address these needs within vehicular IoT systems [2]. While these schemes aim
to enhance privacy and operational efficiency, they often rely on trusted third parties to
maintain high security levels. This reliance creates vulnerability: if the trusted third party is
compromised, the entire security framework can be undermined. Therefore, there is an
urgent need for authentication and key agreement solutions that do not depend on such third
parties. These solutions must ensure the integrity, confidentiality, availability, and
reliability of communications while accommodating the resource constraints inherent in
many vehicular IoT devices.

The smart grid aims to deliver a reliable, sustainable, stable, and efficient electricity supply,
with EV charging management being a critical component of this vision [3]. As the demand
for EV charging services grows, addressing this need efficiently becomes paramount.
Traditional smart grid systems that offer charging services often rely on third parties, which
introduces vulnerabilities: if these third parties are compromised, users may lose access to
essential EV charging services. Moreover, smart grid systems integrated with IoT devices
must balance efficiency with the limited power and memory constraints of EV sensors like
on-board units (OBUs). To overcome these security and efficiency challenges, recent
studies have explored the potential of blockchain technology. Blockchain offers
decentralization, verification, and data integrity, making it a promising solution for various
fields including smart grids, healthcare, finance, and voting [4].

In blockchain systems, data is organized into blocks that contain transactions, with each
transaction linking to previous ones through cryptographic hash functions [5]. Early
blockchain implementations, such as Bitcoin and Ethereum, faced scalability issues,
prompting the development of Hyperledger frameworks. Hyperledgersare designed to
address these scalability challenges without involving cryptocurrency. Building on this
foundation, a blockchain-based security model for EV charging management that utilizes
smart contracts and the lightning network to enhance both security and efficiency within
smart grid systems is proposed [6]. However, their model exhibits several inefficiencies,
including the deposit problem, transaction generation issues, and transaction fees, and it
lacks a guarantee of key security. To address these limitations, an improved secure charging
system for EVs based on Hyper ledger technology is proposed [7].

The need for robust authentication and key agreement schemes in vehicular IoT systems
has been a focus of numerous studies over the past decades [8]. Although these schemes
aim to ensure privacy and improve efficiency, they often rely on trusted third parties to
maintain high security levels. This reliance introduces vulnerabilities, such as susceptibility
to distributed denial of service (DDoS) attacks and privileged insider threats. If the trusted
third party is compromised, the entire security framework can fail. Therefore, it is essential
to develop authentication and key agreement schemes that do not depend on trusted third
parties, ensuring integrity, confidentiality, availability, and reliability, especially in
resource-constrained environments.

312
 Integrating Block chain in EV Charging Systems for Secure and Efficient Infrastructure

II Related Works:

Several studies have explored the application of blockchain technology in the energy
Internet architecture [9]. Authors examined the use of consortium blockchain for energy
trading within industrial IoT environments, employing the Stackelberg game to ensure safe,
fast, and reliable transactions. The concept of automating complex workflows in the energy
IoT sector through smart contracts is introduced [10].

In addition to energy trading, research has increasingly focused on blockchain-based shared

charging systems [11]. a blockchain system is developed that facilitates EV and charging
station interactions without relying on a trusted third party, utilizing smart contracts for
transactions. A consortium block chain to manage charging and discharging transactions
between EVs, establishing a local aggregator as a service node, though without addressing
user transactions across different aggregators is discussed [12]. Numerous studies also apply
smart contracts to enhance shared charging systems. Decentralized security model that
leverages the lightning network and smart contracts to bolster transaction security between
EVs and charging points is explored [13]. A charging station deploy a smart meter equipped
with a blockchain node to monitor power usage and interact with the blockchain through a
smart contract communication module is proposed [14]. Several previous studies have
focused on incentive mechanisms to enhance the quality and efficiency of charging services
[15]. Various incentive strategies used in smart grids and applied a contract-theoretic
approach to the energy trading process, aiming to optimize both service quality and
operational efficiency is summarized.

A scheme utilizing a progressive two-price auction game to address large-scale EV charging

cooperation, ensuring incentive compatibility within a constrained range is developed [16].
An evaluation mechanism for charging services that assesses charging stations based on
user-provided credibility ratings and the endorsement experience values of charging station
nodes is introduced [17].

III Propose Work:

A. Background:

The rapid adoption of Electric Vehicles (EVs) has led to a significant increase in the
deployment of EV charging stations worldwide. As these systems collect vast amounts of
data, including personal information, vehicle details, and payment information, robust data
security mechanisms have become paramount. Traditional data management systems often
face challenges related to data breaches, tampering, and unauthorized access. Blockchain
technology, with its decentralized and immutable nature, offers a promising solution to
these challenges, ensuring the security, integrity, and authenticity of IoT data in EV
charging management.

B. Proposed Goals:

Utilizing block chain technology, Charge Guard aims to establish a secure platform for
managing IoT data within EV charging systems. The proposed objectives are:

313
International Journal of Research and Analysis in Science and Engineering

• Authentication: Ensure that the EV and the charging station authenticate each other.
• Authorization: Ensure that the EV is authorized to use the charging station.
• Integrity and Confidentiality: Ensure that the communication between entities is
secure and tamper-proof.

C. Burrows–Abadi–Needham (BAN) Logic to an Electric Vehicle (EV) charging

management system

Burrows–Abadi–Needham (BAN) Logic is a formal logic used to analyze and reason about
the security of authentication protocols. Developed by Michael Burrows, Martín Abadi, and
Roger Needham in the late 1980s, BAN logic helps to verify that a given protocol can
establish certain security properties, such as mutual authentication or key agreement,
between communicating parties. Applying Burrows–Abadi–Needham (BAN) Logic to an
Electric Vehicle (EV) charging management system can help ensure secure communication
between various entities involved, such as the EV, the charging station, and the backend
management system. The different entities involved in the EV charging management system
is EV (E): The electric vehicle, CS (C): The charging station and Backend Management
System (BMS): The central system that manages charging sessions and billing. By applying
BAN Logic, we can formally verify the security properties of the EV charging management
system, ensuring that the protocol is robust against common security threats such as replay
attacks, impersonation, and unauthorized access.

D. Burrows–Abadi–Needham (BAN) Logic protocol for initiating a charging session:

1. E → C: {E_ID, N1}K_EC
2. C → BMS: {E_ID, N1, C_ID}K_CB
3. BMS → C: {N1, Authorization_Token, N2}K_CB
4. C → E: {Authorization_Token, N2}K_EC

Where:

• E_ID: EV's unique identifier.

• C_ID: Charging station's unique identifier.
• N1, N2: Nonces generated by the EV and the BMS respectively.
• K_EC: Shared secret key between EV and charging station.
• K_CB: Shared secret key between charging station and BMS.
• Authorization_Token: Token generated by BMS to authorize the charging session.

1) BAN Logic Analysis:

a) Message 1: E → C: {E_ID, N1}K_EC

• C ⊳ {E_ID, N1}K_EC: The charging station sees the message.

• C |≡ E ↔ C: The charging station believes it shares a key with the EV.
• C |≡ E |≡ N1: The charging station believes that the EV believes N1 is fresh.

314
 Integrating Block chain in EV Charging Systems for Secure and Efficient Infrastructure

b) Message 2: C → BMS: {E_ID, N1, C_ID}K_CB

• BMS ⊳ {E_ID, N1, C_ID}K_CB: The BMS sees the message.

• BMS |≡ C ↔ BMS: The BMS believes it shares a key with the charging station.
• BMS |≡ C |≡ N1: The BMS believes that the charging station believes N1 is fresh.

c) Message 3: BMS → C: {N1, Authorization_Token, N2}K_CB

• C ⊳ {N1, Authorization_Token, N2}K_CB: The charging station sees the message.

• C |≡ BMS ↔ C: The charging station believes it shares a key with the BMS.
• C |≡ BMS |≡ N1: The charging station believes that the BMS believes N1 is fresh.
• C |≡ BMS |≡ N2: The charging station believes that the BMS believes N2 is fresh.
• C |≡ BMS |≡ Authorization_Token: The charging station believes that the BMS
believes the authorization token is valid.

d) Message 4: C → E: {Authorization_Token, N2}K_EC

• E ⊳ {Authorization_Token, N2}K_EC: The EV sees the message.

• E |≡ C ↔ E: The EV believes it shares a key with the charging station.
• E |≡ C |≡ N2: The EV believes that the charging station believes N2 is fresh.
• E |≡ C |≡ Authorization_Token: The EV believes that the charging station believes
the authorization token is valid.

The proposed protocol with BAN Logic ensures the following

Mutual Authentication: The EV and the charging station can authenticate each other using
the shared key and nonces.

Authorization: The EV receives an authorization token from the BMS through the charging
station, ensuring that only authorized EVs can use the charging station.

Integrity and Confidentiality: The use of shared keys and encrypted messages ensures
that the communication is secure and tamper-proof.

IV Conclusion:

With the growing adoption of electric vehicles (EVs) and the critical need for secure and
efficient charging infrastructures, ensuring the security of communication protocols within
EV charging management systems is paramount. By applying Burrows–Abadi–Needham
(BAN) Logic to the EV charging management system, we have demonstrated that the
protocol effectively ensures mutual authentication, authorization, and the integrity and
confidentiality of communications between the EV, the charging station, and the backend
management system. The analysis confirms that the entities can securely authenticate each
other, authorize charging sessions, and protect the exchanged messages against tampering
and unauthorized access. Consequently, BAN Logic proves to be a valuable tool in verifying
the robustness and security of protocols within EV charging infrastructures.

315
International Journal of Research and Analysis in Science and Engineering

References:

1. N. V. A. Ravikumar, R. S. S. Nuvvula, P. P. Kumar, N. H. Haroon, U. D. Butkar and

A. Siddiqui. "Integration of Electric Vehicles, Renewable Energy Sources, and IoT for
Sustainable Transportation and Energy Management: A Comprehensive Review and
Future Prospects," 12th IEEE International Conference on Renewable Energy Research
and Applications (ICRERA), Oshawa, ON, Canada, pp. 505-511, 2023
2. Li, L., Fan, X., Zhi, B. et al. Highly secure authentication and key agreement protocol
for the internet of vehicles. Telecommun Syst, 2024.
3. Kabeyi, M. J. B., & Olanrewaju, O. A. “Smart grid technologies and application in the
sustainable energy transition: a review”, International Journal of Sustainable Energy,
vol42, Issue1, pp.685–758, 2023
4. Deepak, & Gulia, Preeti & Gill, Nasib & Yahya, Mohammad & Gupta, Punit & Shukla,
Prashant & Shukla, Piyush, "Exploring the Potential of Blockchain Technology in an
IoT-Enabled Environment: A Review," in IEEE Access, vol. 12, pp. 31197-31227,
2024.
5. Bela Shrimali, Hiren B. Patel, Blockchain, “state-of-the-art: architecture, use cases,
consensus, challenges and opportunities”, Journal of King Saud University - Computer
and Information Sciences, vol 34, Issue 9, pp. 6793-6807,2022.
6. X. Huang, C. Xu, P. Wang and H. Liu, "LNSC: A Security Model for Electric Vehicle
and Charging Pile Management Based on Blockchain Ecosystem," in IEEE Access, vol.
6, pp. 13565-13574, 2018.
7. Kim M, Park K, Yu S, Lee J, Park Y, Lee SW, Chung B. “A Secure Charging System
for Electric Vehicles Based on Blockchain”, Sensors (Basel). vol 19, Issue 13 pp.3028,
2019.
8. Siddhant Thapliyal, Mohammad Wazid, D.P. Singh, Ashok Kumar Das, SK Hafizul
Islam, “Robust authenticated key agreement protocol for internet of vehicles-
envisioned intelligent transportation system, Journal of Systems Architecture, Volume
142,2023
9. Z. Tian et al., "Real-Time Charging Station Recommendation System for Electric-
Vehicle Taxis," in IEEE Transactions on Intelligent Transportation Systems, vol. 17,
no. 11, pp. 3098-3109, Nov. 2016
10. W. Tang and Y. J. Zhang, "A Model Predictive Control Approach for Low-Complexity
Electric Vehicle Charging Scheduling: Optimality and Scalability," in IEEE
Transactions on Power Systems, vol. 32, no. 2, pp. 1050-1063, March 2017.
11. Yunhua He, Cui Zhang, Bin Wu, Ziye Geng, Ke Xiao, Hong Li, A trusted architecture
for EV shared charging based on block chain technology, High-Confidence Computing,
Volume 1, Issue 2,2021.
12. S. Aggarwal and N. Kumar, "A Consortium Block Chain-Based Energy Trading for
Demand Response Management in Vehicle-to-Grid," in IEEE Transactions on
Vehicular Technology, vol. 70, no. 9, pp. 9480-9494, Sept. 2021
13. S. Hussain, S. Tahir, A. Masood and H. Tahir, "Blockchain-enabled Secure
Communication Framework for Enhancing Trust and Access Control in the Internet of
Vehicles (IoV)," in IEEE Access, pp1-1, 2019.
14. Y. Wang, Z. Su and K. Zhang, "A Secure Private Charging Pile Sharing Scheme with
Electric Vehicles in Energy Blockchain," 13th IEEE International Conference On Big
Data Science and Engineering (TrustCom/BigDataSE), Rotorua, New Zealand, pp. 648-
654, 2019.

316
 Integrating Block chain in EV Charging Systems for Secure and Efficient Infrastructure

15. Y. Guo, J. Tong and C. Feng, "A Measurement Study of Bitcoin Lightning Network,"
2019 IEEE International Conference on Blockchain (Blockchain), Atlanta, GA, USA,
2019, pp. 202-211, 2019.
16. Mik, E. “Smart contracts: terminology, technical limitations and real world complexity.
Law, Innovation and Technology, vol 9, Issue 2, 269–300, 2017.
17. He L, He J, Zhu L, Huang W, Wang Y, Yu H. Comprehensive evaluation of electric
vehicle charging network under the coupling of traffic network and power grid. PLoS
One. vol17, Issue 9, 2022.

317
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

33. IoT Enabled Infant Incubator for Healthcare

Centers
Sharanappa P. H., Basavaraj M. Angadi,
Mahabaleshwar S. Kakkasageri
Electronics and Communication Engineering Department,
Basaveshwar Engineering College, Bagalkote
Sudha K. S.
Department of Computer Applications (MCA),
Basaveshwar Engineering College, Bagalkote.

ABSTRACT:

Newborns are particularly vulnerable to rough environments and dust and extreme
temperatures can pose life-threatening risks. To address these challenges, we've developed
a baby incubator that simulates the temperature and environmental conditions of a mother's
womb while also monitoring vital medical conditions such as heart rate, skin temperature,
and internal temperature. In response, integrating Internet of Things (IoT) technology into
hospital equipment, like this baby incubator, has become a priority. The goal of the paper
is to provide an app that enables remote monitoring of a baby's condition, allowing doctors
to stay informed and respond quickly if needed. One of the incubator's key features is its
ability to maintain optimal humidity levels, preventing the baby's skin from losing too much
moisture and becoming brittle or cracked. The incubator is also equipped with monitoring
devices that track vital signs, such as temperature and heart rate, enabling healthcare
providers to continuously assess the baby's health.

KEYWORDS:

Internet of Things, Incubator, Health monitoring, Heartbeat.

I Introduction:

A transport incubator is a portable device used to safely transfer sick or premature babies,
such as from a smaller hospital to a larger facility with a Neonatal Intensive-Care Unit
(NICU). In biology, incubators maintain optimal conditions, including temperature,
humidity, and gas levels, for growing cultures. Premature babies, born before 37 weeks,
often have underdeveloped organs like the lungs and digestive tract. Incubators provide the
necessary environment for these babies to survive and thrive by regulating temperature,
318
 IoT Enabled Infant Incubator for Healthcare Centers

humidity, and protection from infections, allergens, excessive noise, and light. They can
adjust temperature automatically based on the baby's needs and may include special lights
for treating neonatal jaundice, ensuring the infant's skin integrity and overall well-being.

We are currently living in an era dominated by smart technologies, often referred to as

"ubiquitous computing" or "Web 3.0." The Internet of Things (IoT) has emerged as a key
area in this technological landscape, complementing other technologies like cloud
computing. Originally discussed in the ITU Internet Reports series, which began in 1997
under the title "Challenges to the Network," the concept of IoT was first coined by Kevin
Ashton in 1999 and officially termed "Internet of Things" in 2005.

Kevin Ashton's vision for IoT involved enabling networked devices to share information
about physical world objects via the web. IoT allows objects to identify themselves and
exhibit intelligent behavior by making or facilitating decisions based on the information
they can share. These objects can either access data collected by other devices or contribute
to other services. With IoT, any object can connect to the internet at any time and from any
place, providing a wide range of services through any network to anyone. This concept has
led to the development of new applications such as smart vehicles and smart homes, offering
services like notifications, security, energy saving, automation, communication, computing,
and entertainment.

Unlike a simple hospital bed, the incubator is designed to create a controlled environment
that supports the health and development of newborns. It regulates critical factors like
oxygen levels, light exposure, and humidity to match the conditions of a mother's womb.
Additionally, the incubator offers protection from external threats such as allergens, loud
noises, bacteria, and viruses. To summarize, our smart baby incubator provides a safe and
controlled environment for newborns, with the added benefit of IoT integration for remote
monitoring. This innovation ensures that healthcare professionals can maintain a high level
of care while minimizing physical contact, making it an essential tool in today's healthcare
landscape.

The rest of the article is structured as follows: Section II reviews related works; Section III
presents the proposed model; Section IV focuses on the analysis of results; and Section V
concludes the paper.

II Related Works:

Temperature regulation is crucial for both living organisms and some semiconductor
materials. The work in [1] aims to control temperature variations in specific applications,
such as baby incubators. Incubators are essential for improving infant survival by providing
a warm environment and minimizing heat loss. It uses Arduino and temperature sensors to
monitor and control the temperature, maintaining it between 36.5-37.2°C, similar to a
mother's womb. The Arduino's programming code is used to achieve precise temperature
control, ensuring the optimal conditions for the baby's health and development.

The authors in [2] presents a design for a central real-time monitoring system for premature
baby incubators, focusing on environmental temperature control. The incubators collect
data using temperature and humidity sensors, as well as web cameras.
319
International Journal of Research and Analysis in Science and Engineering

This data is displayed on a central monitoring interface through networking technology,

allowing for real-time and centralized monitoring of the environment and physical state of
multiple premature babies. This system aims to reduce the risk of medical incidents and
support the healthy growth of premature infants. The need for sophisticated control systems
in incubators is growing quickly in an effort to lower the rates of infant mortality [3]. Many
characteristics in an incubator need to be kept an eye on in order to guarantee the baby's
health. This study describes a sophisticated control system that keeps an eye on and
regulates a number of critical factors that have an impact on a baby's health. Four
temperature sensors are used by the system to control the incubator's temperature and track
the baby's skin temperature. Two sensors are also used to measure humidity. An application
page is made to be easily viewed by users. The technology, which is based on Arduino,
enables precise and seamless operation of the incubator using a serial port.

The work referenced in [4], uses an agent-based technique to provide an intelligent strategy
for IoT data collecting. The main contribution of the proposed research project is to develop
and construct an intelligent system for IoT data gathering. This work involved the
development and evaluation of a fully operational incubator with accurate temperature,
humidity, and airflow control. To test and fine-tune the Mamdani fuzzy logic controller, a
heuristic simulation was developed concurrently with the incubator's development [5].
Ensuring the correct operation of IoT services depends heavily on the security and accuracy
of sensor data transferred across the network. Validating data collected in remote IoT
networks is therefore a problem that is becoming more and more important. Although the
quick fix of adding duplicate identical systems can give validation, real-world change limits
frequently make this difficult or even impossible. Thus, authors have presented an
intelligent validation approach using multi-agents in this study [6] [7].

An incubator is made with such care and attention to detail that it can offer a newborn baby
a secure and healthy environment in which to sleep while its important organs are still
developing. Our incubator can be compared to a basic hospital bed in that it offers precisely
the right quantity of light exposure, humidity, and oxygen levels and all of which are
necessary for a mother's womb [8]. The purpose of the article in [9] is to design an infant
incubator based on Android that a health professional can access and operate over the
Internet using an Android application. The technology is capable of gathering
environmental data from incubators and storing it on a web server. In isolated locations, the
system may have a major impact on lowering the premature infant mortality rate. The work
in [10] proposes a system for infant incubators that uses a set of weight sensors, temperature
and humidity sensors, and a collection of sensors for monitoring the baby's development.
Every incubator equipped with this technology is linked to a central Long Range Networks
(LoRa) network, enabling the registration of medical data in a database. Last but not least,
the system features a Near Field Communication (NFC) interface that enables physician
authentication, tablet-based patient evolution viewing, and the addition of new data.

Larger and smaller divisions make up the chamber. The bigger compartment houses the
baby's mattress, while the smaller compartment houses the temperature and humidity
control device. Relative sensors, fans, lights, heaters, and Arduino Uno microcontrollers are
some of the parts of the control system. A software application written in C has been
designed for implementation [11].

320
 IoT Enabled Infant Incubator for Healthcare Centers

It is a sophisticated take on traditional incubator systems, fully microcontroller based, and

inexpensively developed locally. The system has the ability to regulate the temperature
between 36 and 37 degrees Celsius and the relative humidity between 70% and 75%.

There are already sophisticated incubators on the market, but they are far too costly to serve
the needs of developing countries. The design in [12] aims at creating a low-cost,
thermodynamically sophisticated incubator that can run in a resource-constrained setting. It
includes three novelties: (1) a disposable baby chamber to lower infant death from
nosocomial infections; (2) a passive cooling system with inexpensive heat pipes and
evaporative cooling from clay pots found nearby; and (3) insulated panels and a water-filled
thermal bank that efficiently retain and store heat.

III Proposed Model for Incubator:

The controller, which is the main electrical component in the baby incubator, is what gives
the actuators their electrical commands by receiving signals from the sensors and activating
the actuators in response. The Arduino UNO chip microcontroller, which is widely used
and user-friendly, is the controller utilized in this instance. Since it is simple to program,
this chip is also available for use with the Arduino Micro. It is imperative to measure the
heartbeat pulses first in order to compare the results to the standard values shown in the
below figure. An LDR and a high intensity type LED are used to detect heartbeats. The
LED and LDR are separated by the finger. A photo transistor or photo diode can be used as
a sensor. Transmitted or reflected light can be used to illuminate the skin in visible (red)
light for detecting purposes. The minute variations in transmittance or reflectivity resulting
from the fluctuating blood concentration in human tissue are essentially imperceptible.
Disturbance signals from different sources of noise can have amplitudes that are comparable
to or greater than the amplitude of the pulse signal.

A pulse measurement to be valid, the raw signal must undergo significant preprocessing.
With the novel signal processing method that is being introduced here, disturbance signals
can be effectively suppressed by combining analog and digital signal processing in a way
that allows both to be kept simple. In this configuration, an LDR serves as the detector and
a red LED provides transmitted light illumination. Other illumination and detection
techniques might be employed with the same hardware and software with just minor
modifications to the preamplifier circuit. The DHT11 sensor is made up of a thermistor for
temperature sensing and a capacitive humidity sensor. The humidity detecting capacitor
consists of two electrodes separated by a substrate that can retain moisture as a dielectric.
Changes in humidity levels cause changes in the capacitance value. The resistance values
are measured, processed, and converted into digital form by the IC. This sensor measures
temperature using a negative temperature coefficient thermistor, whose resistance value
decreases as temperature rises.

A. Block Diagram:

Here in this section, the block diagram for the proposed model is discussed. Figure.1 shows
the detailed block diagram for the same. An incubator is a device that keeps an eye on and
maintains surroundings that are healthy for a newborn. It is applied to sick full-term
newborns as well as premature births. The incubator keeps an eye on pressure and oxygen
321
International Journal of Research and Analysis in Science and Engineering

replenishment. In addition, it keeps an eye on the surrounding temperature, radiation, pulse

activity, and air humidity. We employ sensors and data transmission devices in smart
incubators to store data and move it to cloud storage. They can examine the medical data
on their computers and mobile devices from anywhere, and they may act on it from there.
Wi-Fi and infrared technologies, which measure the critical parameters that need to be
regulated for preemies, are the foundation of the design. Variations in this outcome were
reported with prompt alert messages to the patient's household and the appropriate hospital
administration. This research presents a potentially highly helpful biomedical system that
allows physicians to monitor a patient's state from their chair, allowing for prompt and
appropriate patient care. They are able to shield the infant from issues. The suggested
system's block diagram is displayed in the figure. for putting it into practice using a
NodeMCU controller. It is made up of sensors for gas, light, temperature, humidity, and
pulse. Here, cloud storage and a Wi-Fi network have been employed to store medical data.
in order for computers and mobile devices to be able to view the data. In order to prevent
the newborns from unfavorable health conditions, we can inspect the incubator immediately
if there is a problem with the medical data.

Figure 1: Block Diagram

B. Experimental Setup:

The preterm incubator's fundamental physical design consists of a glass container that is set
atop a steel trolley for mobility as shown in figure 2. To ensure the baby's comfort and well-
being, a bed is positioned within the glass container. This allows for the best possible baby
handling because it's not too long for someone of a modest height to use well, nor is it too
short to need the operator to stoop.

322
 IoT Enabled Infant Incubator for Healthcare Centers

Figure 2: Experimental Setup

C. Working:

The Arduino, the brains behind our invention, is primarily responsible for the basic
operation of the Smart Baby Incubator. The glass case is opened, and the infant is settled
onto a cozy bed. The temperature and humidity sensors identify the necessary values as
soon as the system is turned on, and they send them to the microcontroller as shown. The
heater and humidifier are controlled by the microprocessor, which serves as a gateway and
only permits operation when certain humidity and temperature levels are reached. The
diagram below explains the Smart Baby incubator's fundamental operation. After the user
makes adjustments to the temperature and humidity levels, the Arduino compares the
computed values with the user-adjusted values and turns on or off the heater and humidifier
based on the result. The regulated characteristics and the uncontrollable features are the two
primary categories of this incubator's features. The temperature and humidity controls are
the main features of this incubator. In contrast, the camera that tracks the baby's movements
and the heartbeat sensor that displays the baby's heartbeat on the LCD and the Android
mobile application are features that are visible but beyond our control.

IV Results and Discussion:

The experimental and simulated results of the proposed work are discussed in this section.
Figure 3 and 4 shows the sensing of the heart beat and display of the pulse rate respectively.

Figure 3: Heart Beat Sensing

323
International Journal of Research and Analysis in Science and Engineering

Figure 4: Display of Pulse Rate

Figure 5 depicts the simulation results of the proposed work in which the parameters such
as temperature, humidity and heart beat are displayed using blynk application.

Figure 5: Simulation Results

V Conclusion:

The design of a smart baby incubator represents a significant advancement in neonatal care,
providing a safe and controlled environment crucial for the survival and healthy
development of infants, especially those born prematurely. By integrating precise control
systems for temperature, humidity, and airflow, the incubator ensures optimal conditions
that mimic a mother's womb. The use of advanced technologies, such as IoT and fuzzy logic
controllers, allows for real-time monitoring and remote access, enhancing the ability of
healthcare providers to respond swiftly to any changes in the infant's condition.
Additionally, the smart incubator's capabilities to monitor multiple vital parameters
simultaneously reduce the risk of medical complications, safeguarding the well-being of the
newborns. Overall, this innovative approach not only improves the quality of care provided
in neonatal intensive care units but also offers peace of mind to healthcare professionals and
parents alike.

References:

1. M. Koli, P. Ladge, B. Prasad, R. Boria and N. J. Balur, "Intelligent Baby Incubator,"

2018 Second International Conference on Electronics, Communication and Aerospace
Technology (ICECA), Coimbatore, India, 2018, pp. 1036-1042,

324
 IoT Enabled Infant Incubator for Healthcare Centers

doi: 10.1109/ICECA.2018.8474763.
2. M. Zhao, H. Duan, W. Huang and K. Sun, "Central Real-time Monitoring System for
Premature Baby Incubator," 2020 IEEE 3rd International Conference of Safe
Production and Informatization (IICSPI), Chongqing City, China, 2020, pp. 385-390,
doi: 10.1109/IICSPI51290.2020.9332206.
3. M. Ali, M. Abdelwahab, S. Awadekreim and S. Abdalla, "Development of a Monitoring
and Control System of Infant Incubator," 2018 International Conference on Computer,
Control, Electrical, and Electronics Engineering (ICCCEEE), Khartoum, Sudan, 2018,
pp. 1-4, doi: 10.1109/ICCCEEE.2018.8515785
4. P. H. Sharanappa and M. S. Kakkasageri, "Intelligent Information Gathering Scheme in
Internet of Things (IoT)," 2019 11th International Conference on Advanced Computing
(ICoAC), Chennai, India, 2019, pp. 134-138,
doi: 10.1109/ICoAC48765.2019.247128.
5. B. Zimmer, A. A. P. Inks, N. Clark and C. Sendi, "Design, Control, and Simulation of
a Neonatal Incubator," 2020 42nd Annual International Conference of the IEEE
Engineering in Medicine & Biology Society (EMBC), Montreal, QC, Canada, 2020, pp.
6018-6023, doi: 10.1109/EMBC44109.2020.9175407.
6. P. H. Sharanappa and M. S. Kakkasageri, "Multiagent based Information Validation
Scheme in Internet of Things" 2021 International Conference on Smart Generation
Computing, Communication and Networking (SMART GENCON), Pune, India, 2021,
pp. 1-5, doi: 10.1109/SMARTGENCON51891.2021.9645776.
7. P. H. Sharanappa and M. S. Kakkasageri, "Intelligent Agents Assisted Information
Aggregation in Internet of Things," 2022 IEEE Fourth International Conference on
Advances in Electronics, Computers and Communications (ICAECC), Bengaluru,
India, 2022, pp. 1-7, doi: 10.1109/ICAECC54045.2022.9716588
8. Ahmed, A. Sufyan, S. Hussain, A. Rehman, S. Amin and Y. Salam, "Smart Baby
Incubator," 2021 International Conference on Computing, Electronic and Electrical
Engineering (ICE Cube), Quetta, Pakistan, 2021, pp. 1-5,
doi: 10.1109/ICECube53880.2021.9628245.
9. S. Inam, M. F. Qureshi, F. Amin, M. Akmal and M. Z. Rehman, "Android based Internet
Accessible Infant Incubator," 2019 8th International Conference on Information and
Communication Technologies (ICICT), Karachi, Pakistan, 2019, pp. 25-29,
doi: 10.1109/ICICT47744.2019.9001985.
10. S. Sendra, P. Romero-Díaz, J. Navarro-Ortiz and J. Lloret, "Smart Infant Incubator
Based on LoRa Networks," 2018 IEEE/ACS 15th International Conference on
Computer Systems and Applications (AICCSA), Aqaba, Jordan, 2018, pp. 1-6, doi:
10.1109/AICCSA.2018.8612863.
11. Elahi, Fazle & Nisha, Zinat Ara. (2014). Low-Cost Neonatal Incubator with Smart
Control System. 10.13140/2.1.4591.5201.
12. Tran K, Gibson A, Wong D, Tilahun D, Selock N, Good T, Ram G, Tolosa L, Tolosa
M, Kostov Y, Woo HC, Frizzell M, Fulda V, Gopinath R, Prasad JS, Sudarshan H,
Venkatesan A, Kumar VS, Shylaja N, Rao G. Designing a Low-Cost Multifunctional
Infant Incubator. J Lab Autom. 2014 Jun;19(3):332-7.
doi: 10.1177/2211068214530391.

325
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

34. Machine Learning Approaches for Data Storage

in IoT: A Review
Supriya B. Harlapur, Mahabaleshwar S. Kakkasageri
Department of Electronics and Communication Engineering,
Basaveshwar Engineering College, Bagalkote.

ABSTRACT:

The widespread adoption of the Internet of Things (IoT) has resulted in the production of
vast quantities of data originating from many sources. Efficient classification and storage
of this data are critical for deriving actionable insights and enabling real-time decision-
making. An overview of data management in IoT is given in this review paper. It discusses
data storage types including machine learning and deep learning approaches, and
addresses the issues of data storage. Additionally, it highlights how to address these
challenges through data classification, anomaly detection, predictive maintenance, and
enhancing the compression of data and de-duplication. By leveraging advanced storage
architectures and machine learning techniques, effective, safe, and scalable IoT data
storage systems can be developed to meet the growing demands of the IoT ecosystem.

KEYWORDS:

IoT, data storage, machine learning, data classification.

I Introduction:

The IoT represents a transformative model in which commonplace objectsare

interconnected through the internet, enabling them to collect, exchange, and process data
autonomously. This interconnectivity has led to an exponential increase in data generation,
posing significant challenges for data storage solutions. An effective way to manage the
data lifecycle requirements of a system is through various designs, processes, and
methodologies that are together referred to as data management. The massive volume of
data and its unique features [1] using conventional database systems wouldn't be a superior
choice. Indeed, a number of concepts form the foundation of the design of IoT data
management systems. Numerous data management strategies, including middleware-based
IoT focused on data and sources, data storage and indexing solutions, and IoT data scheme
support solutions, have been presented based on these various ideas. A middleman between
items and data storage places is offered by middleware-based IoT techniques.

326
 Machine Learning Approaches for Data Storage in IoT: A Review

As a result, the middleware makes data stream routing possible. IoT data can be produced
very quickly, in large volumes, and with a variety of data kinds. Data storage solutions
provide to these possible issues. Efficient data storage is crucial in IoT environments due to
the unique characteristics of IoT data, including its high volume, velocity, variety, and
veracity [2]. Traditional data storage methods often fall short in addressing these challenges,
necessitating the exploration of more advanced and tailored storage architectures.

Because there is a wide range of data providers and users, IoT data have unique properties
such as temperature sensors, humidity sensors, cameras, body sensors, and RFID readers.
These data are produced by billions of connected objects across numerous industries, such
as supply chain management, healthcare, the military, and transportation either continuously
or at discrete intervals [3]. The classification of IoT data includes:

A. Heterogeneous and Multi-Source Characteristics:

• RFID Data: Used for tracking and identifying in a variety of applications, including
supply chain management, road tolling, passports, and animal tracking. Data is sent by
radio waves by RFID tags, which can be active or passive.
• Sensor Data: Produced by Wireless Sensor Networks to monitor and control
phenomena such as weather, temperature, noise, and video.
• Positioning Data: GPS or local positioning systems, such as Wi-Fi access points and
cellular base stations, are used to locate tagged objects. This data is crucial for both
static and mobile IoT objects.
• Metadata: Describes data, enabling users to find and access relevant information about
objects, processes, and systems, maximizing data sharing.

B. Large Scale Characteristics:

• The number of IoT objects is expected to reach 212 billion globally by the end of 2020,
generating massive amounts of data. Efficient data storage mechanisms, such as cloud
computing, data canters, and fog computing, are necessary to handle this scale.

C. Spatio-temporal Characteristics:

• IoT data reflects the current state of the environment or phenomenon but is collected at
discrete times, leading to spatio-temporal characteristics that need to be considered for
accurate real-time analysis.

D. Multi-dimensional Characteristics:

• IoT applications monitor various indicators like weather, temperature, noise, humidity,
light, and pressure, leading to multi-dimensional data. Efficient techniques are required
to manage this complexity, especially when data must be captured continuously, at
regular intervals, or upon request.

327
International Journal of Research and Analysis in Science and Engineering

E. Interoperability Characteristics:

• Data-sharing facilitates collaborative work between different IoT applications. For

example, in the Internet of Medical Things (IoMT), combining physiological data with
traffic information can improve emergency response times and first-aid planning.

F. Contextual Characteristics:

• Context includes location, network conditions, type of service, and Quality of Service
(QoS). It describes the environment in which IoT data is generated or consumed, the
available resources, and the expected performance metrics like latency.

These characteristics underscore the need for robust and adaptable data management
solutions to effectively harness the potential of IoT systems. This review paper aims to
provide a comprehensive analysis of various data storage methods suitable for IoT,
including cloud, edge, and hybrid storage models. By understanding the landscape of IoT
data storage, we can better address the demands of modern IoT systems and leverage new
technologies to optimize data management and utilization. Our paper is organised as
follows: section I gives brief introduction about the work. Related work is discussed in
section II. The section III and IV represents data storage and data storage challenges
respectively and section V highlights various machine learning approaches. The section VI
concludes the paper.

II Related Work:

The current state of data management schemes, layer-based architecture using distributed
approach, efficient IoT data storage and query processing mechanisms for satisfying IoT
application are discussed in [1-3]. Author proposed compression data solutions using
discrete cosine transform solutions [4] and dividing the data into multiple fragments, and
encrypting the data using owner's private key. The Coordinate-based Indexing (COIN)
mechanism for the data sharing in edge computing and maintains a virtual space where the
switches and the data indexes [5]. An online client algorithm based on machine learning
algorithm for IoT unstructured big data analysis using the online data entered by the
customer to implement background data mining, the parallel way to verify its efficiency
through machine learning algorithms such as K-nearest neighbor algorithm [6]. The
proposed work presents [7] threshold secret sharing scheme for storing aggregate data in
IoTs in cloud storage. Data storage and verification of local repair code shading block chain
based on bilinear accumulator is presented in [8]. The data storage process is performed by
using a dual block chain topology that includes a lightweight block chain (local block chain)
and a public block chain. The proposed Service Data Processing Mechanism (SDPM) [9,
10] aims to improve data storage efficiency within the context of a Service-oriented
Architecture underlying the Internet of Things (So AIoT). The hybrid framework with a
software and hardware integration strategy for an industrial platform that exploits features
from a Relational Database (RDB) and Triple store and Synchronization mechanism is
discussed in [11]. The state-of-the-art technologies and research trends concerning RSBD
storage and computing is discussed in [12].

328
 Machine Learning Approaches for Data Storage in IoT: A Review

Cloud-based optimized remote sensing data storage technology and cloud storage
technology represented by OSS, New SQL, and NoSQL techniques and solution for RSBD
storage and management described in [13].Hadoop deep learning architecture and
implementation process are analyzed and software performance testing tool Load Runner
discussed in [14].Map Chain-D is designed for practical IIoT applications with storage,
latency, and communication constraints are discussed in[15].ID-based data storage scheme
utilizing anonymous key generation in fog computing distributed data storage method with
the combined K-means and PSO clustering mechanism organized with the binary decision
tree C4.5 in the IoT area with considering efficiency and reliability approaches discussed in
[16, 17].Improved RSA accumulator to solve the problem of data storage expansion in the
block chain [18].Two NoSQL databases Google Big table and Dynamo based on CAP
theorem [19].The Hadoop distributed file system (HDFS)cloud storage approach using
Heuristic algorithm is presented in [20].Edge KV, a decentralized storage system designed
for the network edge using the Yahoo! Cloud Serving Benchmark (YCSB) to analyze the
system’s performance under realistic workloads is proposed in [21].Optimized hash
algorithm of IoT data access storage in cloud computing[22].A Gated Recurrent Unit based
Digital Twin Framework for Data Allocation and storage in IoT-enabled Smart Home
Networks. Low-priority data is stored and processed at the Macro-based Stations (MBSs),
and high-priority data is transferred to the upper [23]. A block chain storage optimization
scheme based on RS erasure code is proposed in [24]. Novel encoding scheme [25] called
Pathed Dewey Order and a two-layer mapping method to store XML documents in HBase
tables. Hadoop based big data secure storage scheme using a homomorphic encryption
algorithm, dual-threaded encrypted storage is presented in [26]. A real-time intelligent
monitoring and notification system (RT-IMNS) in cold storage using Artificial Neural
Network (ANN) is proposed in [27].The database heterogeneity of the USPIOT platform an
XML based unified management mechanism database is designed in [28].Complex
Perceptual Data Placement Algorithm Terminal Node Load Balancing Strategy and the
cost-minimization storage selection using two heuristic algorithms: Dynamic Programming
(DP) based algorithm and Greedy Style (GS) algorithm, for optimizing the choice of data
storage based on IoT application service requirements are described in [29, 30]. The paper
[31] propose an algorithm to design the data model from ontological information of the
domain and a set of most frequent queries expected to run on the database. Data storing
using Ethereum block chain and IPFS and develops smart contracts and Ethereum D Appare
presented [32]. Qualitative and quantitative analysis of time series encoding algorithms
regarding to various data features. The comparison is conducted in Apache IoTDB, an open
source time series database[33].Adaptive Multi-Model Middle-Out using reinforcement
learning based compression for time-series data[34].The optimized data storage models the
Extreme Learning Machine, SVM-based optimized model and Elastic Chain model for
synthetic data are mentioned in [35].The Inter Planetary File System (IPFS), a decentralized
storage solution that employs Message Queuing Telemetry Transport (MQTT) protocol for
efficient Content Identifiers (CIDs) transfer and a database for archiving CID values and
associated metadata [36].

III Data Storage:

In the context of the IoT, data storage techniques refer to the methods used to store and
manage the vast amounts of data generated by IoT devices. As IoT devices often produce a

329
International Journal of Research and Analysis in Science and Engineering

continuous stream of data, efficient and scalable storage solutions are necessary to handle
this data effectively. Here are some common data storage types used in IoT:

1. Cloud Storage: Storing IoT data in the cloud is a popular option due to its scalability,
accessibility, and cost-effectiveness. Cloud storage providers offer large-scale,
distributed storage systems that can handle massive amounts of data. IoT devices can
transmit data to the cloud in real-time, where it is stored, processed, and made available
for further analysis.
2. Edge Storage: Edge storage involves storing IoT data locally on the edge devices
themselves, closer to the source of data generation. This approach helps reduce latency
and bandwidth requirements by processing and storing data locally. Edge storage is
particularly useful in scenarios where real-time data analysis or quick response times
are crucial.
3. Hybrid storage(Edge-Cloud): A hybrid storage system combines both edge and cloud
storage to create a more flexible and efficient data storage solution. Sensors and local
gateways process real-time data, such as traffic conditions or air quality, to make
immediate decisions or send alerts, Aggregated data from all sensors is periodically sent
to the cloud for long-term storage, comprehensive analysis, and machine learning to
detect patterns and trends.
4. Distributed Storage: Distributed file systems are designed to store and manage data
across multiple nodes or devices in a distributed network. These systems enable
efficient storage and retrieval of large amounts of data, ensuring data availability, fault
tolerance, and scalability. Examples of distributed file systems used in IoT include
Hadoop Distributed File System (HDFS) and Google File System (GFS).
5. Database Systems: Database systems are commonly used for structured storage and
efficient data retrieval in IoT applications. Relational databases (e.g., MySQL,
PostgreSQL) and NoSQL databases (e.g., MongoDB, Cassandra) are utilized to
organize, store, and query IoT data based on specific requirements. These databases
provide mechanisms for indexing, querying, and aggregating data, enabling efficient
data processing and analysis.
6. Time-Series Databases: Time-series databases (TSDBs) are designed specifically for
handling time-stamped data, which is prevalent in IoT applications. TSDBs optimize
the storage and retrieval of time-series data, making them ideal for storing sensor data,
telemetry data, and other time-dependent IoT data. Examples of TSDBs include Influx
DB and Prometheus.
7. Spatial data storage: involves managing and querying data that has a geographic or
spatial component, such as locations, shapes, and boundaries. This type of data is crucial
in various applications, including geographic information systems (GIS), location-
based services, urban planning, and environmental monitoring.
8. Hierarchical data storage: is a way of organizing data into a tree-like structure that
represents relationships between data elements in a parent-child hierarchy. This
structure is often visualized as a tree with branches, where each node represents a data
element, and the connections between nodes represent the relationships between them
9. Object Storage: Object storage is a method of storing unstructured data as objects,
each having a unique identifier. It provides a highly scalable and cost-effective solution
for storing large volumes of IoT data. Object storage systems, such as Amazon S3 and
OpenStack Swift, are commonly used in IoT deployments for storing multimedia data,
logs, and files.

330
 Machine Learning Approaches for Data Storage in IoT: A Review

10. Multi-modal data storage refers to a storage architecture designed to handle and
manage different types or modalities of data within a unified system. Unlike traditional
storage systems that may focus on a single data type (e.g., text, images, or time-series
data), multi-modal data storage integrates various data types into a cohesive framework.
This approach is particularly useful in environments where diverse data types need to
be processed and analyzed together, such as in modern applications that involve text,
images, videos, sensor data, and more.

It's important to note that different IoT applications and use cases have varying data storage
requirements, and a combination of these techniques may be employed to meet specific
needs. Factors such as data volume, velocity, variety, and latency requirements play a role
in selecting the appropriate data storage approach for an IoT system.

IV Data Storage Challenges:

Data storage in IoT environments presents several challenges due to the unique
characteristics of IoT systems and the massive volume of data they generate. Here are some
of the main challenges:

1. Volume and Velocity of Data: IoT devices generate vast amounts of data continuously.
Handling this high volume and high-velocity data can overwhelm traditional storage
systems.
2. Data Heterogeneity: IoT data comes in various formats (structured, semi-structured,
and unstructured) from different devices and sensors. Managing and storing this
heterogeneous data efficiently is challenging.
3. Data Integrity and Quality: Ensuring the accuracy, consistency, and reliability of data
from diverse sources can be difficult. Inconsistent or erroneous data can lead to
incorrect analytics and decision-making.
4. Latency and Real-Time Processing: Many IoT applications require real-time data
processing and low-latency responses. Traditional storage solutions may not meet these
requirements, necessitating the use of edge computing or hybrid storage solutions.
5. Scalability: Storage systems must scale seamlessly to accommodate the growing
number of IoT devices and the increasing volume of data they generate. This requires
scalable architecture and infrastructure.
6. Security and Privacy: IoT data often contains sensitive information. Ensuring the
security and privacy of data during storage and transmission is critical. This includes
protecting against unauthorized access, data breaches, and ensuring compliance with
regulations.
7. Energy Efficiency: IoT devices and the storage systems they interact with must be
energy-efficient. High energy consumption can be a limiting factor, especially for
battery-operated devices and remote deployments.
8. Cost Management: Storing massive amounts of data can be expensive. Finding cost-
effective storage solutions while maintaining performance and reliability is a significant
challenge.
9. Data Lifespan and Retention: Determining how long to store IoT data and managing
data lifecycle policies (from creation to deletion) can be complex, especially when
different data types have different retention requirements.

331
International Journal of Research and Analysis in Science and Engineering

10. Network Bandwidth: Transferring large amounts of data from IoT devices to
centralized storage can strain network bandwidth, especially in environments with
limited connectivity. Efficient data transmission and compression techniques are
needed.
11. Integration and Interoperability: IoT systems often involve multiple vendors and
platforms. Ensuring seamless integration and interoperability between different storage
systems and IoT devices can be challenging.

To overcome data storage challenges in IoT, future directions should focus on enhancing
edge computing for real-time processing, optimizing hybrid storage solutions that combine
edge and cloud resources, and implementing advanced data compression and de-duplication
techniques.

Integrating AI and machine learning can enhance data management, while block chain
technology can ensure data integrity and security. Developing energy-efficient storage
solutions, promoting interoperability standards, designing scalable architectures, and
ensuring data privacy and regulatory compliance are also crucial. Additionally, exploring
quantum computing could revolutionize data processing and storage, providing
unprecedented capabilities for the growing IoT ecosystem.

V Machine Learning Approaches:

Machine learning approaches for IoT data storage enhance efficiency and effectiveness
through various techniques. Hierarchical deep learning models, such as CNNs and RNNs,
classify data at multiple levels, while transfer learning leverages pre-trained models for
faster classification.

Data storage is optimized with collaborative filtering, predicting and storing only relevant
data, and model pruning and quantization, reducing model sizes for edge storage. Hybrid
storage solutions like edge-cloud collaboration balance latency and resource usage, and
federated learning trains models across edge devices, enhancing privacy.

Anomaly detection with auto encoders and SVMs prioritizes critical data, and predictive
maintenance using LSTM networks and regression models forecasts storage needs. Future
directions focus on adaptive learning systems, energy-efficient algorithms, and enhanced
security and privacy, ensuring robust and scalable IoT data management. The Table 1
depicts about different machine learning approaches based on various data storage types.

Machine learning offers innovative solutions to address the challenges of IoT data storage.
By leveraging advanced classification techniques, optimizing storage allocation, and
employing hybrid storage models, ML can significantly enhance the efficiency and
effectiveness of IoT data management. Future research should focus on developing
adaptive, energy-efficient, and secure ML algorithms to further improve IoT data storage
solutions.

332
 Machine Learning Approaches for Data Storage in IoT: A Review

Table 1: Different Machine learning approaches for data storage types

Data Storage Types

Cloud Edge Hybrid Distrib Time- Spatial Hierar Multi
storage Storage Storage uted Series Data chical Modal
storag storage Storage Data Data
e Storage Storag
e
k-Nearest Reinforc Hierarc Distrib Recurre Convolut Hierarc Multi-
Neighbors ement hical uted nt ional hical modal
(k-NN) Learning Deep Machi Neural Neural Clusteri Neural
(e.g., Q- Learnin ne Networ Network ng Networ
[6] Learning g Learni ks s (CNNs) ks
, DQN) Models ng (RNNs)
[14] Frame [27]
works
Machine Learning Algorithms

[20]
Random Lightwei Collabo Graph Long Geospati Recursi Attenti
Forest ght rative NeuralShort- al ve on
neural Filterin NetworTerm Analysis Neural Mecha
network g ks Memory Tool Networ nisms
(GNNs Networ ks
) ks
(LSTMs
Gradiet Online Transfe Bayesi Tempor DBSCA Decisio Auto
Boosting learning r an al N n Trees encode
Machines algorith Learnin Networ Convolu r
(GBM) m [6] g ks tional
Networ
ks
(TCNs)
Support K- Model Federat ARIMA Geograp Tree- Genera
vector nearest Pruning ed (Auto hically based tive
Machines neighbor and Learni Regressi Weighte Ensemb Advers
s[17] Quantiz ng ve d le arial
ation Integrat Regressi Method Networ
ed on s like, ks
Moving (GWR) Gradien (GANs
Average t )
) Boostin
g Trees)

333
International Journal of Research and Analysis in Science and Engineering

VI Conclusion:

The ongoing field of IoT presents numerous data storage challenges, including managing
vast and heterogeneous data, ensuring real-time processing, maintaining data integrity, and
addressing security and scalability concerns.

Traditional storage solutions often fall short in meeting these demands, necessitating
innovative approaches such as hybrid storage architectures that combine the strengths of
edge and cloud storage. Furthermore, machine learning techniques offer promising
solutions for optimizing data management, from classification and anomaly detection to
predictive maintenance and data compression.

By embracing these advanced methodologies, we can develop robust, efficient, and secure
data storage systems that are capable of supporting the expansive and dynamic nature of
IoT applications.

References:

1. Jerome Oswald Ebenezer, Calduwel Newton, “Data Management in IoT: A Detailed

Survey”, International Journal of Information Technology Research and Applications,
Vol. 2, June 2023: pp 18-32
2. Diene, J. P. C. Rodrigues, et al., “Data Management Techniques for Internet of Things”,
MSSP 138 106564, Elsevier,2020 pp.1-18
3. Diallo, Ousmane, Philippe Dhaussy, and Ciprian Teodorov. "Data management
mechanisms for internet of things: A position paper." In 2019 International Conference
on Computational Science and Computational Intelligence (CSCI). IEEE, 2019, pp.
1214-1219.
4. Subhash Rathod, Mangesh D. Salunke, Meghna Yashwante, et al. Ensuring Optimized
Storage with Data Confidentiality and Privacy Preserving for Secure Data Sharing
Model Over Cloud”, International Journal of International Journal of Intelligent
Systems and Applications in Engineering IJISAE, 2023, pp 35–44 |
5. Junjie Xie, Chen Qian, Deke Guo, Minmei Wang, Shouqian Shi, Honghui Chen
“Efficient Indexing Mechanism for Unstructured Data Sharing Systems in Edge
Computing,” In Proc. IEEE INFOCOM 2019 - IEEE Conference on Computer
Communications, 29 April 2019 - 02 May 2019.
6. Rui Hou1, YanQiang, Kong, Bing Cai, Huan Liu “Unstructured big data analysis
algorithm and simulation of Internet of Things based on machine learning,” In journal
ofNeural Computing and Applications, Vol. 32, 2020, pp. 5399–5407
7. Bypour, H., M. Farhadi, and RJIJoE Mortazavi. "An efficient secret sharing-based
storage system for cloud-based internet of things." International Journal of
Engineering 32, vol no. 8 2019: pp:1117-1125.
8. Ren, Yongjun, Xinyu Liu, Pradip Kumar Sharma, Osama Alfarraj, Amr Tolba,
Shenqing Wang, and Jin Wang. "Data storage mechanism of industrial IoT based on
LRC sharding blockchain." Scientific Reports 13, vol no. 1 2023: 2746.
9. Barron, Alfredo, Dante D. Sanchez-Gallegos, Diana Carrizales-Espinoza, J. L.
Gonzalez-Compean, and Miguel Morales-Sandoval. "On the efficient delivery and
334
 Machine Learning Approaches for Data Storage in IoT: A Review

storage of IoT data in edge–fog–cloud environments." Sensors 22, vol no. 18 2022:
7016.
10. Wang, Shu-Ching, Wei-Ling Lin, Mao-Lun Chiang, and Shin-Jie Wang. "Improve Data
Storage Efficiency with Service-oriented Architecture underlying the Internet of
Things." International Journal of Applied Science and Engineering 17, vol no. 2 2020:
pp:191-206.
11. Grevenitis, K., F. Psarommatis, A. Reina, W. Xu, I. Tourkogiorgis, J. Milenkovic, J.
Cassina, and D. Kiritsis. "A hybrid framework for industrial data storage and
exploitation." Procedia CIRP 81 2019 pp: 892-897.
12. Xu, Chen, Xiaoping Du, Xiangtao Fan, Gregory Giuliani, Zhongyang Hu, Wei Wang,
Jie Liu et al. "Cloud-based storage and computing for remote sensing big data: a
technical review." International Journal of Digital Earth 15, vol no. 1 2022: pp:1417-
1445.
13. Hendawi, Abdeltawab, Jayant Gupta, Liu Jiayi, Ankur Teredesai, Ramakrishnan
Naveen, Shah Mohak, and Mohamed Ali. "Distributed NoSQL data stores: Performance
analysis and a case study." IEEE International Conference on Big Data (Big Data),
IEEE, 2018.pp. 1937-1944.
14. Duan, Yuntao, Jiangdai Li, Gautam Srivastava, and Jyh-Haw Yeh. "Data storage
security for the internet of things." The Journal of Supercomputing, Vol. 76, no. 11,
pp:8529-8547.
15. Wu, Tiantong, Guillaume Jourjon, Kanchana Thilakarathna, and Phee Lep Yeoh.
"MapChain-D: A distributed blockchain for IIoT data storage and
communications." IEEE Transactions on Industrial Informatics 19, vol no. 9 2023, pp:
9766-9776.
16. Lin, Han-Yu, Tung-Tso Tsai, Pei-Yih Ting, and Ching-Chung Chen. "An improved ID-
based data storage scheme for fog-enabled IoT environments." Sensors 22, vol 11 2022:
4223.
17. Taheri, Negar, Shahram Jamali, and Mohammad Esmaeili. "Achieving Perform ability
and Reliability of Data Storage in the Internet of Things." International Journal of
Engineering and Manufacturing 12, no. 1 2022: 12.
18. Wang, Jin, Wencheng Chen, Lei Wang, Yongjun Ren, and R. Simon Sherratt.
"Blockchain-based data storage mechanism for industrial internet of things." Intelligent
Automation and Soft Computing 26, vol no. 5 2020, pp: 1157-1172.
19. Thant, Phyo Thandar, and Thinn Thu Naing. "Improving the availability of NoSQL
databases for Cloud Storage." PhD diss., MERAL Portal, 2011.
20. Lin, Jenn-Wei, Joseph M. Arul, and Jia-Ting Kao. "A bottom-up tree-based storage
approach for efficient IoT data analytics in cloud systems." Journal of Grid
Computing 19 2021, pp: 1-19.
21. Sonbol, Karim, Öznur Özkasap, Ibrahim Al-Oqily, and Moayad Aloqaily. "EdgeKV:
Decentralized, scalable, and consistent storage for the edge." Journal of Parallel and
Distributed Computing 144 2020, pp: 28-40.
22. Wang, Mingzhe, and Qiuliang Zhang. "Optimized data storage algorithm of IoT based
on cloud computing in distributed system." Computer Communications 157 2020, pp:
124-131.
23. Singh, Sushil Kumar, Manish Kumar, Sudeep Tanwar, and Jong Hyuk Park. "GRU-
based digital twin framework for data allocation and storage in IoT-enabled smart home
networks." Future Generation Computer Systems 153 2024, pp: 391-402.

335
International Journal of Research and Analysis in Science and Engineering

24. Li, Chunlin, Jing Zhang, Xianmin Yang, and Luo Youlong. "Lightweight block chain
consensus mechanism and storage optimization for resource-constrained IoT
devices." Information Processing & Management 58, no. 4 2021, vol: 102602.
25. Bao, Liang, Jin Yang, Chase Q. Wu, Haiyang Qi, Xin Zhang, and Shunda Cai.
"XML2HBase: Storing and querying large collections of XML documents using a
NoSQL database system." Journal of Parallel and Distributed Computing 161 2022,
pp: 83-99.
26. Guan, Shaopeng, Conghui Zhang, Yilin Wang, and Wenqing Liu. "Hadoop-based
secure storage solution for big data in cloud computing environment." Digital
Communications and Networks 10, no. 1 2024, pp: 227-236.
27. Afreen, Hina, and Imran Sarwar Bajwa. "An IoT-based real-time intelligent monitoring
and notification system of cold storage." IEEE Access 9 2021, pp: 38236-38253.
28. Qi, Weiqiang, Hongkai Wang, and Tong Chen. "Multimedia System Design and Data
Storage Optimization Based on Machine Learning Algorithm." Computational
Intelligence and Neuroscience 2022, no. 1 2022, pp: 6426551.
29. Wu, Jiansheng, Weimin Xu, and Jiarong Xia. "Load balancing cloud storage data
distribution strategy of internet of things terminal nodes considering access
cost." Computational Intelligence and Neuroscience 2022, pp: 7849726.
30. Danish, Syed Muhammad, Kaiwen Zhang, and Hans-Arno Jacobsen. "BlockAM: An
adaptive middleware for intelligent data storage selection for Internet of Things.”
International Conference on Decentralized Applications and Infrastructures (DAPPS),
IEEE, 2020: pp. 61-71.
31. Sen, Poly Sil, and Nandini Mukherjee. "Designing a NoSQL Database for Efficient
Storage and Retrieval of Health Data." In 8th International Conference on Future
Internet of Things and Cloud (FiCloud), IEEE, 2021, pp. 262-269.
32. Ye, Hyoeun, and Sejin Park. "Reliable vehicle data storage using blockchain and
IPFS." Electronics 10, no. 10 2021, pp: 1130.
33. Xiao, Jinzhao, Yuxiang Huang, Changyu Hu, Shaoxu Song, Xiangdong Huang, and
Jianmin Wang. "Time series data encoding for efficient storage: A comparative analysis
in apache iotdb." Proceedings of the VLDB Endowment 15, no. 10 2022, pp: 2148-2160.
34. Yu, Xinyang, Yanqing Peng, Feifei Li, Sheng Wang, Xiaowei Shen, Huijun Mai, and
Yue Xie. "Two-level data compression using machine learning in time series database."
In 36th International Conference on Data Engineering (ICDE),2020, pp: 1333-1344
35. Jia, Dayu, Junchang Xin, Zhiqiong Wang, and Guoren Wang. "Optimized data storage
method for sharding-based blockchain." IEEE Access 9 2021, pp: 67890-67900.
36. Lee, Chanhyuk, Jisoo Kim, Heedong Ko, and Byounghyun Yoo. "Addressing IoT
storage constraints: A hybrid architecture for decentralized data storage and centralized
management." Internet of Things 252024: 101014.

336
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

35. Performance Analys is of Inter-Satellite Optical

Wireless Communication
Onkar A., Anudeep Daggupati,
Shreevatsa Kulkarni, Vineeth Pelliyembil
SENSE,
Vellore Institute of Technology, Vellore, India.
Abstract:

When it comes to data transmission technologies, laser frequency offers a significant

advantage over other conventional means like radio wave sand microwaves. It is more
suited for usage in bidirectional space communication with high-speed objects and
communication channels due to its fast data rate, small antenna size for both the transmitter
and receiver, and high through put. This work aims to provide a brief over view of the
application of lasers in inter satellite communication systems and to compare several
metrics, such as eye diagrams, BER (Bit Error Rates), Q-factors, etc., utilizing MIMO,
SISO, SIMO, and MISO over a certain distance. We will offer a condensed comparative
analysis and some recommendations for further study.

Keywords:

BER, SINR, MIMO, Q-factor, Modulator.

I. Introduction:

Since space communication makes it possible to send data and information over great
distances, it is essential to modern civilization. For a long time, conventional space
communication methods like microwave transmission have been in use.

But thanks to technological developments, a brand-new kind of space communication called

laser communication has surfaced. When compared to more conventional communication
methods like microwaves, laser communication technology has a number of advantages.

First off, the increased data rate of laser communication is one of its main benefits. Large
volumes of data may be transferred quickly by the usage of laser technology, which enables
greater transmission speeds. Second, when it comes to total security and confidentiality,
laser communication out performs conventional modes of communication.

337
International Journal of Research and Analysis in Science and Engineering

This is because laser shave small be am divergence, which makes it harder for eaves
droppers to jam or intercept the communication. Furthermore, laser communication is more
robust against interference, which increases its depend ability in difficult circumstances.

Thirdly, there is a chance for higher capacity with laser communication technology. Larger
capacity is a benefit of laser communication over conventional microwave transmission. As
a result, communication systems can transmit more data at once and operate more
efficiently.

While there are many benefits to laser communication, there are also some difficulties and
things to consider. One of these is laser beam tracking and positioning with extreme
precision. This is required to prevent signal loss or degradation due to even small
misalignments, and to provide precise and dependable communication between satellites.
Furthermore, as environmental circumstances might have an impaction signal propagation,
it is important to carefully evaluate these factors while using laser technology for space
communication.

II Related Work:

It is difficult to achieve dependable and effective communication in optical wireless

communication systems because of things like atmospheric effects, signal intensity changes,
and interference.

This highlights how crucial it is to choose the best system architecture and modulation style
for depend able communication in these kinds of systems. It is impossible to ignore how
Wave Division Multiplexing (WDM) affects the performance of Is OWC (Inter-satellite
Optical Wireless Communication).

The performance and potential of Is OWC systems are improved by the inclusion of Dense
Wave Division Multiplexing (DWDM). The LP-IsOWC system can further improve system
performance and reliability by reducing BER by using Erbium Dropped Fiber Amplifiers
(EDFAs) as booster amplifiers.

Overall, the selected papers highlight how crucial it is to take into account a number of
variables while developing and refining optical wireless communication systems, including
modulation format, system architecture, and the possible advantages of WDM integration.
Engineers can efficiently build and improve the performance of optical wireless
communication systems by accounting for these aspects.

III Methodology:

We mostly use RZ, NRZ and Manchester line coding scheme to convert our data in binary
input. We are sending this binary input code to optical modulator, which is used to convert
the information carried by an electric current into beam of light. This beam of light is sent
to the receiver where the data processing is happening.

338
 Performance Analys is of Inter-satellite Optical Wireless Communication

IV. Systemmodel:

Figure 1: Procedural flow

We are working on free space communication, so according to space we need to prepare

our channel, where we can use different channel conditions like wavelength, range,
attenuation in free space. The beam of light sent by sender is received by a photo receiver
at the receiver side.

We have finalized to go with the factors namely:

Q-factor which represents the ability of the system to maintain its operating frequency, Eye
diagram represents the quality of the transmitted signal BER provides the ability of system
to correctly transmit the data bit. From these factors we will check the performance of the
system. and compare them with the existing data.

339
International Journal of Research and Analysis in Science and Engineering

A. Input Data: For the proposed model input is generated by Pseudo Random bit sequence
generator and converted this to NRZ input format.
B. Modulation Schemes: The input data is sent to MZ modulator for modulation and this
data will be attached to the Laser for communication. The modulation scheme used in
this system is PSK (Phase Shift Keying).
C. Channel Conditions: We must use free space for this model so the conditions will
mostly be constant. There are several channel conditions like Attenuation, Turbulence
and path loss which are related the dimensions of the channel.
D. Output: The transmitted data is received by the Photo-receptor which will be converted
to electrical signal and will be analyzed by the low pass filter.

V. Implementation:

We are comparing different Line coding schemes like RZ, NRZ and Manchester coding.
Out of the se NRZ is giving better results for all the communication techniques. We are
implementing four different communication techniques like MIMO, MISO, SIMO, and
SISO. The parameters we are using are:

Parameters Values
Wavelength 850 nm
Range 45000Km
Bit Rate 1.8Gbps
Power of Laser 41Watts

A. SISO (Single Input Single Output):

In this model, SISO(Single Input Single Output) is implemented, NRZ pulse generator is
used to generate the input. The signal gets modulated in Analytical modulator and then gets
into the OWC(Optical wireless communication) channel. Then it goes to low pass b
Bettlefilter and the output is got in BER analyzer through 3Rregenerator.

340
 Performance Analys is of Inter-satellite Optical Wireless Communication

B. SIMO (Single Input Multiple Output):

In this model, only a single input is connected through NRZ generator and CW laser. Power
splitter is present to split the signals into multiple parts. Here the power is split into four
parts hence 4 OWC channels are used. Hence 4 APDs, 4 Low pass filters, 43 R regenerators
and 4 BER analyzers used as shown in the figure.

C. MISO (Multiple Input Single Output):

In this model, 4 NRZ inputs are given with four modulators. Then the modulated signals go
through 4 OWC channels. Then there is a power combiner being used and optical amplifiers
as shown in figure which combines the power and amplifies output, respectively. Only one
output is got from the APD, Low pass filter, 3RregeneratorandBERanalyzer.

341
International Journal of Research and Analysis in Science and Engineering

D. MIMO (Multiple Input Multiple Output):

From the diagram, it is observed that four different inputs are given to power combiner
which send that signal through the OWC channel. A power splitter is used to divide the
modulated signals into four different signals which are received by an APD and converted
into electrical signal. A Low pass Bessel filters, 3Rregenerators and BER analyzers are used
to generate the results for our understanding.

VI. Result and Discussion:

A. Results for SISO:

342
 Performance Analys is of Inter-satellite Optical Wireless Communication

B. Results for SIMO:

C. Results for MISO:

D. Results for MIMO:

343
International Journal of Research and Analysis in Science and Engineering

SISO SIMO MISO MIMO

Q-Factor 8.95911 4.79084 1.50661 9.16926
BER 1.54215e-19 7.75124e-07 0.0430995 2.32373e-20

For execution of this system the Opti-System software is used through which the above
results are generated. By comparing all the diagrams, the table is created to check the
results. From the above table, an inference can be taken that when Q-factor, BER of different
communication systems are compared, the Q factor of MIMO is the best when compared
to other systems. From above diagrams we can conclude that both Q-factor and BER are
inversely proportional to each other.

VI. Conclusion:

We can conclude that MIMO system works more efficiently than other systems with the Q-
factor of 9.16926. Like MIMO, SIMO also works efficiently which is having Q- factor of
4.79084, so it can be inferred that the proposed system works more efficiently for multiple
out puts. When SISO and MISO are compared, single out put is obtained and it is inferred
that SISO works effectively with Q-factor of 8.95911, where as the Q-factor for MISO is
1.50661 which is much lesser than the other systems.

When it comes to Bit Error Rate (BER) it can be observed that MIMO system is having
lowest value of 2.32373e-20 and MISO has the highest value of 0.0430995. From this it
can be concluded that the Q- factor and BER are inversely proportional to each other that
means the system having minimum BER and maximum Q- factor is working more
efficiently.

VII. References:

1. Mohamed Mohsen Tawfik, Mohamed Fathy AboSree, Mohamed Abaza, and Hussein
Hamed Mahmoud Ghouz “Performance Analysis and Evaluation of Inter-Satellite
Optical Wireless Communication System (Is OWC) from GEO to LEO at Range
45000km,” in IEEEPHOTONICSJOURNAL, VOL.13, NO. 4, AUGUST 2021.
2. Yogendra Singh et al. / International Journal of Computer Science & Engineering
Technology (IJCSET), “Performance Analysis of Optical Wireless Communication
Channel Link at Various Bit Rates” ISSN: 2229-3345 Vol. 5 No.01 Jan2014.
3. Tomoaki Ohtsuki, “Performance Analysis of Optical Wireless MIMO with Optical
Beat Interference,” (C) 2005IEEE.
4. Bijila Susan Viju1, Asha R S2, Almaria Joseph3, “Performance Comparison of 10gbps
Inter Satellite Optical Wireless Communication System with Different Pulse
Generators”, © 2019 JETIR May 2019, Volume 6, Issue5.

344
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

36. Review of Structural Health Monitoring

Methods: A Machine Learning Approach
Rashmi M. Kittali
Department of Electronics and Communication Engineering,
BGMIT, Karnataka, India
Mudhol, Karnataka, India.
Ashok V. Sutagundar
Department of Electronics and Communication Engineering,
Basaveshwar Engineering College,
Bagalkote, Karnataka, India.

ABSTRACT:

The discipline of structural health monitoring (SHM) has benefited from a great deal of
novel sensing and monitoring systems based on machine vision-based technologies during
the last 20 years. Some of the technology's unique intrinsic benefits include immunity to
electromagnetic interference, extended range, high accuracy, noncontact, nondestructive,
and broad-spectrum, multi-target surveillance. Numerous techniques for structural
condition inspection and structural dynamic assessment based on machine vision have been
put forth. Measurements of the physical characteristics of the structure, such as
displacement, strain/stress, rotation, vibration, fracture, and spalling, are also made in
real-world settings. This review article's goal is to provide an overview of the fundamental
ideas and real-world uses of the machine vision-based technology used in structural
monitoring. It also aims to integrate the technology with other contemporary sensing
techniques and address systematic error causes.

I Introduction:

The majority of the public infrastructure in today's society consists of concrete constructions
like pavements or bridges and tunnels made of various materials such as concrete, asphalt,
or other stone types. The demand for maintenance rises with greater use and ageing
facilities, and improper maintenance can result in poor conditions or structural defects. This
is a typical issue. For instance, a recent assessment on the health of the US infrastructure by
the American Society of Civil Engineers claims that over 2000 dams and over 46,000
bridges are structurally defective, and that on average, every fifth mile of roadway is in bad
condition [1].

345
International Journal of Research and Analysis in Science and Engineering

Alongside those maintenance procedures, the process of structural health monitoring

(SHM) and assessment is often conducted. This is a crucial stage since it considers the
overall health of the structures and prioritizes and catalogues any potential irregularities.
However, because it is frequently done at the target place, this procedure is labor-intensive.
It can also be risky at times because infrastructure that are difficult to reach may not be able
to be used as planned, such as bridges, tunnels, or highways that must be closed to the
public. Furthermore, because the procedure is frequently completed by hand, human
variables are involved, which might have a detrimental impact on the results of such
structural assessments. Amid the SHM process, the objective is to discover markers for
current or future harm. A few common harm sorts that show up on structures are surface
splits on asphalts, concrete, bricks, stone and black-top, spalling of concrete and broken or
eroded steel [8]. The most common defect of concern is crack which spoils the overall
integrity of the structure. The methods used to detect might be as straightforward as visual
inspections or taking pictures or videos [5], as well as employing lidar scanners to create
3D representations [9], [10], and ultrasonic wave exams [11], [12].

It is evident that once the data has been collected using such methods, anomaly detection
algorithms may be applied and the data can be reviewed at a later time. Cracks frequently
don't show up in a consistent manner; they might be of various sizes and forms, obscured
by moss or leaves, or otherwise partially covered by occlusions. Because some fractures are
so tiny, it may be difficult to tell them apart from their surroundings. Although it might be
challenging for people to operate under these settings, machine learning (ML) offers a
potential remedy. Over the past few decades, machine learning techniques in the field of
computer vision have significantly improved in performance, thanks in part to the
introduction of far more potent hardware and software. Numerous technologies and
applications have been influenced by these developments in machine learning (ML) in
conjunction with computer vision applications in recent years. This can include making
autonomous driving possible [13], [14], outperforming humans in picture categorization
[15], or supporting experts in the medical field using various scans for diagnosis, including
CT [16] or X-rays [17]. Most of the most advanced techniques for such tasks are powered
by deep neural networks, which are a subset of machine learning called deep learning (DL).
This has also made it possible for research into automating some aspects of the surface
health inspection process, such as automatically identifying and measuring fractures as
surface defects, to soar [7], [18] [20]. Figure 1 illustrates the overall procedure for
integrating DL into the structural health monitoring framework for cracks. In recent years,
several researchers have conducted reviews on the subject of structural monitoring and
condition assessment using machine vision. Casciati and Wu [24] provided a brief
introduction to visual positioning systems for structural monitoring. Jiang et al. [25]
outlined the development and application of close-range photogrammetry to measure bridge
deformation and geometry.

Koch et al. [26] presented a comprehensive integration of state-of-the-art defect detection

and condition assessment for concrete and asphalt structures based on computer vision
technologies. However, a comprehensive overview of visual structural monitoring and
condition assessment is still desirable.

In this paper, we provide a detailed assessment of machine vision-based monitoring of civil

engineering infrastructure, including the main approaches and practical applications.

346
 Review of Structural Health Monitoring Methods: A Machine Learning Approach

Figure 1: Machine Learning Way for SHM

Summary of the vision based structural monitoring and condition assessment is still
desirable. This paper provides a detailed assessment of machine vision-based monitoring of
civil engineering infrastructure, including key approaches and practical applications.

II Operation/ Tasks:

Surface crack extraction using computer vision may be divided into many activities. These
tasks coincide with conventional computer vision. The choice of classification, detection,
and segmentation jobs depends on the needed level of detail in the SHM process.

A. Crack Classification:

Assigning the appropriate class label to a picture is the aim of the image classification
problem in the traditional setting of computer vision. To classify images within the crack
space in SHM, the easiest approach is to identify which images have cracks and which ones
don't. Though this assignment is limited to binary classification, it might still prove difficult
because fractures may not always be the only surface imperfection that resembles one.
Building on this two-label classification problem, the crack classification task may be
expanded to infer the type of crack in addition to determining whether a crack is present.

B. Crack Detection:

Crack detection goes beyond categorisation in that it highlights the position of the crack in
order to offer more information in addition to determining whether or not it exists. This can
be advantageous since simple classification tasks just highlight the presence of cracks; they
do not attempt to offer any form of fracture localisation. In contrast, detection tasks seek to
identify break locations. One can approximate the position of a fracture by identifying sub-
regions of a picture and then piecing them back together to create a bigger component.
Bounding box construction is another method of detection. This job has garnered a
significant amount of recent effort and is frequently studied in other large datasets like
COCO [32]. Still, this might not always be the best option for cracks. In the worst scenario,
the bounding box only indicates the outermost points of a fracture, with a significant portion
of the region inside this box being devoid of cracks, because some fissures only manifest as
a single linear structure.

C. Crack Segmentation:

This method solves the problem of not having sufficient information about cracks in
detection and classification step. This job assigns a specified class label to pixels in images
347
International Journal of Research and Analysis in Science and Engineering

or voxels in 3-D volumes. The segmentation problem may be divided into semantic and
instance segmentation. The former classifies all pixels or voxels in input data, without
distinguishing between instances. In a picture, many cracks are labelled as belonging to the
same class. In contrast, instance segmentation distinguishes between instances. Different
cracks are allocated the class label "crack," yet their instance labels differ.

III Review of Machine Vision Based Techniques:

A visual measurement system typically includes image capture equipment (digital camera,
lens, gripper), computer, and image processing software. The image processing software
platform is integrated with appropriate computational techniques to generate machine
parameters for structural monitoring. To use image processing algorithms, an image
containing predefined targets is captured by a digital camera. The targets are monitored
using digital image processing and pattern matching algorithms [27-30], and the structural
displacement at the target points on the structure can be calculated. In this case, horizontal
and vertical displacements, referred to as two-dimensional (2D) displacements, can be
calculated using digital image correlation [31], Mean Shift tracking algorithm [32],
CamShift tracking algorithm [33] and the Lucas-Kanade method [34]. Baska et al. [39] used
two types of cameras to capture images of multiple targets fixed on a railway bridge during
train passage and determined the displacement response using three image processing
techniques: digital image correlation, edge detection, and pattern matching. Li et al. [40]
proposed a pose-graph optimized displacement estimation method to reduce the estimation
error in a visually controlled paired structured light system.

Nayyarloo et al. [41] developed an image processing system to monitor the seismic response
of structures using line-scan cameras. Chang et al. [42] proposed a CCD camera-based
method to measure the vertical displacement of a bridge. Santos et al. [43] performed the
calibration of an image processing system in measuring the structural displacement of a
long-lane suspension bridge. When applying visual methods to measure vibrating targets,
there is a high risk of measurement uncertainty due to motion.

Table 1: Review of Different Learning Algorithms for SHM

Type of Task Explanation Examples

Learning Performed
Classification CNN to classify whether [31], [51]–
C images/patches contain cracks or other [53]
defect types.
Classification of patches using a CNN [54], [55]
D followed by merging to obtain a coarse
Supervised
detection map. [54], [55
CNN that classifies the presence of a crack [39], [52],
within single pixel. Using a sliding window [56]
s
approach, every pixel within an image is
classified. [39], [52], [56]

348
 Review of Structural Health Monitoring Methods: A Machine Learning Approach

Type of Task Explanation Examples

Learning Performed
Approaches using a U-Net shape with [36], [63]–
s popular image classification architectures [65]
as the encoder and a custom decoder
s Approaches that use an ensemble of CNNs [73], [74]
Approaches that use a CNN to predict the [40], [77]
Q
size of a crack
Two-step approaches that first segment a [77]–[80]
Q crack followed by calculating their
dimensions
Two stage approaches in which a CNN [81]
classifier is trained on image labels and its
class activation maps are used to create
S
pseudo segmentation labels of the training
Semi and images. Those pseudo labels are then used
Weakly to train a segmentation algorithm.
Supervised
Approaches that train on classification [55], [82]
S labels and then use thresholding to create a
segmentation map
S Training on coarse segmentation labels [83]
Transformation of an input image into [84], [85]
latent space or frequency domain before
reversing transforming it back into an
Unsupervised S
image. The differences between the input
and output then segment areas belonging to
cracks or other anomalies.

Blur caused by the movement of the camera and target. Motion blur will cause significant
systematic errors and incomplete measurement data, as the target search process may not
achieve accurate detection.

In recent years, research efforts have centred on creating algorithms for mitigation and noise
reduction, as well as ways for interpreting blurred pictures [62-64]. Wang et al. [65]
suggested a vibration measuring approach based on blurred pictures that takes into account
the link between the geometric moments of undistorted and blurred motion. To address the
issue of motion blur in an online particle imaging system for analysing wear particles, Peng
et al. [66] created an image restoration approach to improve the quality of dynamic particle
pictures. Becker [67] conducted a research to analyse motion blur using several fundamental
methodologies and a diverse set of parameters. Wu et al. [68] proposed an image line-by-
line degradation model and a restoration strategy to account for spatially variable
deterioration.

349
International Journal of Research and Analysis in Science and Engineering

Table 2: Review On Literature of ML for SHM

Method Input Output Source

K-means clustering and High quality picture of Area of moss and crack [194]
canny edge detection walls in stone monuments
KNN Color and geometric data Loss of material and [78]
extracted discoloration on walls
Banalization of image Real-time scene image Quantity of dust [79]
processing operations deposited
Deep CNN Pathology analysis Damage analysis [80]
Fuzzy inference system Rebound hammer, fractal Weathering extent [81]
dimensions
Fuzzy logic Dimensions ,texture and Quality of material [82]
fissure properties of stone used for construction
blocks

IV Review of Machine Learning Based Approaches:

Here, the techniques are split as supervised, semi as well as weakly supervised and
unsupervised learning. Table 1 shows the commonly used methods for crack classification
(C), detection (D), segmentation (S) and quantification (Q). Table 2 highlights the possible
input and output combination for various machine learning ways.

V Research Gap:

Here we look into the various research gaps observed potentially after the review of some
literature.

A. Crack Classification:

While multiple datasets for various activities demonstrating fractures are accessible, there
is a scarcity of large-scale publically available datasets. The ImageNet dataset, with its large
sample size, has boosted DL research in the broader computer vision area. As a result, this
area would greatly benefit from having a single, large-scale dataset for training and
evaluating algorithms. While the GAPs v2 subset in [53] and SDNet2018 [74] with over
50k pictures for classification are promising, annotated data for segmentation and
quantification is also needed. To address this issue, consider combining several datasets, as
suggested in [75]. However, the issue of inaccurate labelling persists across several datasets
in this region, as shown in subsection VI-A2. To ensure consistency and accuracy, this data
might benefit from re-labelling by experienced specialists. Unfortunately, this would
require significant expenditures and effort.

350
 Review of Structural Health Monitoring Methods: A Machine Learning Approach

B. Semi, Weakly and Unsupervised Learning:

Semi, weak, and unsupervised learning techniques for cracks are under-represented
compared to supervised learning. To address the previously described dataset concerns, it
may be advantageous to build a standardised dataset for these sorts of learning. Currently,
writers manually adjust the label quality of datasets [71], [73], which may not be the most
efficient strategy for the future. Creating a uniform dataset for several sorts of learning helps
speed up research and improve algorithm comparability. Currently, there are few
unsupervised learning techniques that use deep learning. Research in this area might be
highly beneficial, given the challenges of getting and labelling data.

C. Temporal Data:

Another issue that goes far beyond DL algorithms is the availability of temporal data
showing cracks. Currently available datasets only show cracks at a single point in time.
While experts can use their expertise to determine the severity of a crack and its future
development, datasets containing data showing the evolution of a crack over a specific time
period are not yet available. Continuous data prediction has made significant progress
thanks to DL over other areas such as video sequence prediction [76] and time series
prediction [77]. SHM can benefit greatly from such datasets to support predictive
maintenance actions. DL algorithms can be applied to predict future crack propagation and
determine if and when preventive maintenance actions need to be taken. However, the
limitations are similar to those of generating regular data sets. Moreover, it may take time
for cracks to grow, and it may take years to create such a dataset with training, validation,
and test data.

D. Metric:

The vast amount of work in this field has led to a wide variety of evaluation methods,
datasets, and metrics. Work performing similar tasks in this field would greatly benefit from
consistent evaluation procedures and well-defined metrics, which would greatly increase
the comparability of algorithms. As research interest in crack quantification grows, it may
be beneficial to establish a standardized measure for determining the accuracy of predicted
degrees of rotation inside cracks, including length and thickness.

VI Conclusion and Discussion:

Currently, DL algorithms have achieved state-of-the-art results in various fields and are also
applied to SHM. The paper gives a summary of research and accomplishments in the
domain of structural monitoring of civil infrastructure using machine vision and machine
learning approaches. In particular, we reviewed approaches in different learning types
(supervised, semi-supervised, weakly supervised and unsupervised) along with an overview
of the common metrics and datasets used. Also, overview of the problems facing research
in this field, outline possible research gaps and provide perspectives on future research
directions. As a main result, we identified a wealth of research in the field of supervised
learning, but note that it is difficult to compare architectures and performance due to the
lack of standardized common metrics, datasets and the problem of annotation of datasets.

351
International Journal of Research and Analysis in Science and Engineering

We hope that these issues will be addressed in the future, and believe that further research
in the area of semi-supervised and weakly unsupervised data will advance research and
mitigate the problems associated with small datasets and difficult annotation.

While great advances have been made in vision-based SHM, there are still limitations and
obstacles to overcome. For example, i) most current research is carried out in laboratory
using scale physical models, which may not convert to field continuous monitoring due to
complicated site characteristics. ii) Additionally, the quality of pictures acquired by the
vision system is still an issue. (iii) As an interdisciplinary and cutting-edge technology,
developing a scientific and effective coordination mechanism among civil researchers
remains a key challenge.

References:

1. “COVID-19’s Impacts on America’s Infrastructure,” 2020.

2. A. Homer and G. Wong, “Thousands of Miles of Uk Roads in Poor Condition,”
https://www.bbc.co.uk/news/uk-england-43407167, Mar. 2018.
3. H. Oliveira and P. L. Correia, “Automatic Road Crack Segmentation Using Entropy
and Image Dynamic Thresholding,” in European Signal Processing Conference. IEEE,
2009, pp. 622–626.
4. Y. Pan, X. Zhang, G. Cervone, and L. Yang, “Detection of Asphalt Pavement Potholes
and Cracks Based on the Unmanned Aerial Vehicle Multispectral Imagery,” IEEE
Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 11,
no. 10, pp. 3701– 3712, 2018.
5. S. Sony, S. Laventure, and A. Sadhu, “A Literature Review of NextGeneration Smart
Sensing Technology in Structural Health Monitoring,” Structural Control and Health
Monitoring, vol. 26, no. 3, p. e2321, 2019.
6. C. Yuan, B. Xiong, X. Li, X. Sang, and Q. Kong, “A Novel Intelligent Inspection Robot
with Deep Stereo Vision for Three-Dimensional Concrete Damage Detection and
Quantification,” Structural Health Monitoring, p. 14759217211010238, 2021.
7. K. Jang, Y.-K. An, B. Kim, and S. Cho, “Automated Crack Evaluation of a High-Rise
Bridge Pier Using a Ring-Type Climbing Robot,” Computer-Aided Civil and
Infrastructure Engineering, vol. 36, no. 1, pp. 14–29, 2021.
8. B. F. Spencer Jr, V. Hoskere, and Y. Narazaki, “Advances in Computer Vision-Based
Civil Infrastructure Inspection and Monitoring,” Engineering, vol. 5, no. 2, pp. 199–
222, 2019.
9. B. Riveiro, M. DeJong, and B. Conde, “Automated Processing of Large Point Clouds
for Structural Health Monitoring of Masonry Arch Bridges,” Automation in
Construction, vol. 72, pp. 258–268, 2016.
10. M. Song, S. Yousefianmoghadam, M.-E. Mohammadi, B. Moaveni, A. Stavridis, and
R. L. Wood, “An Application of Finite Element Model Updating for Damage
Assessment of a Two-Story Reinforced Concrete Building and Comparison with
Lidar,” Structural Health Monitoring, vol. 17, no. 5, pp. 1129–1150, 2018.
11. N. K. Mutlib, S. B. Baharom, A. El-Shafie, and M. Z. Nuawi, “Ultrasonic Health
Monitoring in Structural Engineering: Buildings and Bridges,” Structural Control and
Health Monitoring, vol. 23, no. 3, pp. 409–422, 2016.

352
 Review of Structural Health Monitoring Methods: A Machine Learning Approach

12. E. Erdogmus, E. Garcia, A. S. Amiri, and M. Schuller, “A Novel Structural Health

Monitoring Method for Reinforced Concrete Bridge Decks Using Ultrasonic Guided
Waves,” Infrastructures, vol. 5, no. 6, p. 49, 2020.
13. S. Milz, G. Arbeiter, C. Witt, B. Abdallah, and S. Yogamani, “Visual SLAM for
Automated Driving: Exploring the Applications of Deep Learning,” in Conference on
Computer Vision and Pattern Recognition Workshops. IEEE, 2018, pp. 247–257.
14. H. Rashed, E. Mohamed, G. Sistu, V. R. Kumar, C. Eising, A. ElSallab, and S.
Yogamani, “Generalized Object Detection on Fisheye Cameras for Autonomous
Driving: Dataset, Representations and Baseline,” in Winter Conference on Applications
of Computer Vision. IEEE, 2021, pp. 2272–2280.
15. K. He, X. Zhang, S. Ren, and J. Sun, “Delving Deep into Rectifiers: Surpassing Human-
Level Performance on ImageNet Classification,” in International Conference on
Computer Vision. IEEE, 2015, pp. 1026– 1034.
16. L. Zhou, Z. Li, J. Zhou, H. Li, Y. Chen, Y. Huang, D. Xie, L. Zhao, M. Fan, S. Hashmi
et al., “A Rapid, Accurate and Machine-Agnostic Segmentation and Quantification
Method for CT-Based Covid-19 Diagnosis,” IEEE Transactions on Medical Imaging,
vol. 39, no. 8, pp. 2638–2652, 2020.
17. I. M. Baltruschat, H. Nickisch, M. Grass, T. Knopp, and A. Saalbach, “Comparison of
Deep Learning Approaches for Multi-Label Chest XRay Classification,” Scientific
Reports, vol. 9, no. 1, pp. 1–10, 2019.
18. A. Riid, R. Louk, R. Pihlak, A. Tepljakov, and K. Vassiljeva, “Pave- ˜ ment Distress
Detection with Deep Learning Using the Orthoframes Acquired by a Mobile Mapping
System,” Applied Sciences, vol. 9, no. 22, p. 4829, Jan. 2019.
19. R. Louk, A. Riid, R. Pihlak, and A. Tepljakov, “Pavement Defect ˜ Segmentation in
Orthoframes with a Pipeline of Three Convolutional Neural Networks,” Algorithms,
vol. 13, no. 8, p. 198, 2020.
20. D. Li, Q. Xie, X. Gong, Z. Yu, J. Xu, Y. Sun, and J. Wang, “Automatic Defect Detection
of Metro Tunnel Surfaces Using a Vision-Based Inspection System,” Advanced
Engineering Informatics, vol. 47, p. 101206, Jan. 2021
21. R. Khallaf and M. Khallaf, “Classification and analysis of deep learning applications in
construction: A systematic literature review,” Automation in Construction, vol. 129, p.
103760, 2021.
22. C.-Z. Dong and F. N. Catbas, “A Review of Computer Vision–Based Structural Health
Monitoring at Local and Global Levels,” Structural Health Monitoring, p.
1475921720935585, 2020.
23. K. Gopalakrishnan, “Deep Learning in Data-Driven Pavement Image Analysis and
Automated Distress Detection: A Review,” Data, vol. 3, no. 3, p. 28, 2018.
24. W. Cao, Q. Liu, and Z. He, “Review of Pavement Defect Detection Methods,” IEEE
Access, vol. 8, pp. 14 531–14 544, 2020.
25. K. Lu, “Advances in Deep Learning Methods for Pavement Surface Crack Detection
and Identification with Visible Light Visual Images,” arXiv preprint arXiv:2012.14704,
2020.
26. H. Zakeri, F. M. Nejad, and A. Fahimifar, “Image Based Techniques for Crack
Detection, Classification and Quantification in Asphalt Pavement: A Review,”
Archives of Computational Methods in Engineering, vol. 24, no. 4, pp. 935–977, 2017.
27. A. Mohan and S. Poobal, “Crack Detection Using Image Processing: A Critical Review
and Analysis,” Alexandria Engineering Journal, vol. 57, no. 2, pp. 787–798, 2018.
28. H. S. Munawar, A. W. Hammad, A. Haddad, C. A. P. Soares, and S. T. Waller, “Image-
Based Crack Detection Methods: A Review,” Infrastructures, vol. 6, no. 8, p. 115, 2021.
353
International Journal of Research and Analysis in Science and Engineering

29. Z. Du, J. Yuan, F. Xiao, and C. Hettiarachchi, “Application of Image Technology on

Pavement Distress Detection: A Review,” Measurement, p. 109900, 2021.
30. Y.-A. Hsieh and Y. J. Tsai, “Machine Learning for Crack Detection: Review and Model
Performance Comparison,” Journal of Computing in Civil Engineering, vol. 34, no. 5,
p. 04020038, 2020.
31. M. Eisenbach, R. Stricker, D. Seichter, K. Amende, K. Debes, M. Sesselmann, D.
Ebersbach, U. Stoeckert, and H.-M. Gross, “How to Get Pavement Distress Detection
Ready for Deep Learning? A Systematic Approach,” in International Joint Conference
on Neural Networks. IEEE, 2017, pp. 2039–2047.
32. T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollar, and C. L.
Zitnick, “Microsoft COCO: Common Objects in Context,” in European Conference on
Computer Vision. Springer, 2014, pp. 740–755.
33. Q. Zou, Z. Zhang, Q. Li, X. Qi, Q. Wang, and S. Wang, “DeepCrack: Learning
Hierarchical Convolutional Features for Crack Detection,” IEEE Transactions on Image
Processing, vol. 28, no. 3, pp. 1498– 1512, 2018.
34. Y. Liu, J. Yao, X. Lu, R. Xie, and L. Li, “DeepCrack: A Deep Hierarchical Feature
Learning Architecture for Crack Segmentation,” Neurocomputing, vol. 338, pp. 139–
153, 2019
35. F. Yang, L. Zhang, S. Yu, D. Prokhorov, X. Mei, and H. Ling, “Feature Pyramid and
Hierarchical Boosting Network for Pavement Crack Detection,” IEEE Transactions on
Intelligent Transportation Systems, vol. 21, no. 4, pp. 1525–1535, 2019.
36. J. Konig, M. D. Jenkins, M. Mannion, P. Barrie, and G. Morison, ¨ “Optimized Deep
Encoder-Decoder Methods for Crack Segmentation,” Digital Signal Processing, vol.
108, p. 102907, 2021.
37. J.-M. Guo, H. Markoni, and J.-D. Lee, “BARNet: Boundary Aware Refinement
Network for Crack Detection,” IEEE Transactions on Intelligent Transportation
Systems, pp. 1–16, 2021.
38. V. Hoskere, Y. Narazaki, T. A. Hoang, and B. F. Spencer, “MaDnet: Multi-Task
Semantic Segmentation of Multiple Types of Structural Materials and Damage in
Images of Civil Infrastructure,” Journal of Civil Structural Health Monitoring, vol. 10,
no. 5, pp. 757–773, 2020.
39. Y. Inoue and H. Nagayoshi, “Deployment Conscious Automatic Surface Crack
Detection,” in Winter Conference on Applications of Computer Vision. IEEE, 2019,
pp. 686–694.
40. V. P. Tran, T. S. Tran, H. J. Lee, K. D. Kim, J. Baek, and T. T. Nguyen, “One Stage
Detector (RetinaNet)-Based Crack Detection for Asphalt Pavements Considering
Pavement Distresses and Surface Objects,” Journal of Civil Structural Health
Monitoring, Nov. 2020.
41. A. Luxmoore, “Holographic Detection of Cracks in Concrete,” NonDestructive Testing,
vol. 6, no. 5, pp. 258–263, 1973.
42. D. H. Mendelsohn, “Automated Pavement Crack Detection: An Assessment of Leading
Technologies,” in North American Conference on Managing Pavements, vol. 3.
Citeseer, 1987, pp. 297–314. [43] N. Tanaka and K. Uematsu, “A Crack Detection
Method in Road Surface Images Using Morphology,” in Workshop on Machine Vision
Applications. IAPR, 1998, pp. 154–157.
43. L. Peng, W. Chao, L. Shuangmiao, and F. Baocai, “Research on Crack Detection
Method of Airport Runway Based on Twice-Threshold Segmentation,” in International
Conference on Instrumentation and Measurement, Computer, Communication and
Control. IEEE, 2015, pp. 1716–1720.
354
 Review of Structural Health Monitoring Methods: A Machine Learning Approach

44. M. Salman, S. Mathavan, K. Kamal, and M. Rahman, “Pavement Crack Detection using
the Gabor Filter,” in Conference on Intelligent Transportation Systems. IEEE, 2013, pp.
2039–2044.
45. K. C. P. Wang, Q. Li, and W. Gong, “Wavelet-Based Pavement Distress Image Edge
Detection with a Trous Algorithm,” Transportation Research Record, vol. 2024, no. 1,
pp. 73–81, 2007.
46. H. Zhao, G. Qin, and X. Wang, “Improvement of Canny Algorithm Based on Pavement
Edge Detection,” in International Congress on Image and Signal Processing, vol. 2.
IEEE, 2010, pp. 964–967.
47. Y. Shi, L. Cui, Z. Qi, F. Meng, and Z. Chen, “Automatic Road Crack Detection using
Random Structured Forests,” IEEE Transactions on Intelligent Transportation Systems,
vol. 17, no. 12, pp. 3434–3445, 2016.
48. K. Fernandes and L. Ciobanu, “Pavement Pathologies Classification using Graph-
Based Features,” in International Conference on Image Processing. IEEE, 2014, pp.
793–797.
49. A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep
Convolutional Neural Networks,” in Advances in Neural Information Processing
Systems, 2012, pp. 1097–1105.
50. L. Zhang, F. Yang, Y. D. Zhang, and Y. J. Zhu, “Road Crack Detection Using Deep
Convolutional Neural Network,” in International Conference on Image Processing.
IEEE, 2016, pp. 3708–3712.
51. L. Pauly, D. Hogg, R. Fuentes, and H. Peel, “Deeper Networks for Pavement Crack
Detection,” in International Symposium on Automation and Robotics in Construction.
IAARC, 2017, pp. 479–485.
52. R. Stricker, M. Eisenbach, M. Sesselmann, K. Debes, and H.-M. Gross, “Improving
Visual Road Condition Assessment by Extensive Experiments on the Extended GAPs
Dataset.” in International Joint Conference on Neural Networks. International Neural
Network Society, 2019, pp. 1–8.
53. X. Wang and Z. Hu, “Grid-Based Pavement Crack Analysis Using Deep Learning,” in
International Conference on Transportation Information and Safety. IEEE, 2017, pp.
917–924.
54. J. Konig, M. Jenkins, M. Mannion, P. Barrie, and G. Morison, ¨ “A Weakly-Supervised
Surface Crack Segmentation Method Using Localisation with a Classifier and
Thresholding,” arXiv preprint arXiv:2109.00456, 2021.
55. Z. Fan, Y. Wu, J. Lu, and W. Li, “Automatic Pavement Crack Detection Based on
Structured Prediction with the Convolutional Neural Network,” arXiv preprint
arXiv:1802.02208, 2018.
56. J. Cheng, W. Xiong, W. Chen, Y. Gu, and Y. Li, “Pixel-Level Crack Detection Using
U-Net,” in TENCON. IEEE, 2018, pp. 0462–0466.
57. M. D. Jenkins, T. A. Carr, M. I. Iglesias, T. Buggy, and G. Morison, “A Deep
Convolutional Neural Network for Semantic Pixel-Wise Segmentation of Road and
Pavement Surface Cracks,” in European Signal Processing Conference. IEEE, 2018,
pp. 2120–2124.
58. J. Konig, M. D. Jenkins, P. Barrie, M. Mannion, and G. Mori- ¨ son, “A Convolutional
Neural Network for Pavement Surface Crack Segmentation Using Residual
Connections and Attention Gating,” in International Conference on Image Processing.
IEEE, 2019, pp. 1460– 1464.

355
International Journal of Research and Analysis in Science and Engineering

59. R. Augustauskas and A. Lipnickas, “Aggregation of Pixel-Wise U-Net Deep Neural

Networks for Road Pavement Defects Detection,” in Soft Computing for Problem
Solving 2019. Springer, 2020, pp. 109–120.
60. S. L. Lau, E. K. Chong, X. Yang, and X. Wang, “Automated Pavement Crack
Segmentation Using U-Net-Based Convolutional Neural Network,” IEEE Access, vol.
8, pp. 114 892–114 899, 2020.
61. W. Wang and C. Su, “Convolutional Neural Network-Based Pavement Crack
Segmentation Using Pyramid Attention Network,” IEEE Access, vol. 8, pp. 206 548–
206 558, 2020.
62. Z. Qu, W. Chen, S.-Y. Wang, T.-M. Yi, and L. Liu, “A Crack Detection Algorithm for
Concrete Pavement Based on Attention Mechanism and Multi-Features Fusion,” IEEE
Transactions on Intelligent Transportation Systems, 2021.
63. H. Chen and H. Lin, “An Effective Hybrid Atrous Convolutional Network for Pixel-
Level Crack Detection,” IEEE Transactions on Instrumentation and Measurement, vol.
70, pp. 1–12, 2021.
64. C. Han, T. Ma, J. Huyan, X. Huang, and Y. Zhang, “CrackW-Net: A Novel Pavement
Crack Image Segmentation Convolutional Neural Network,” IEEE Transactions on
Intelligent Transportation Systems, 2021.
65. T. Kobayashi, “Spiral-Net with F1-Based Optimization for ImageBased Crack
Detection,” in Asian Conference on Computer Vision. Springer, 2018, pp. 88–104.
66. A. Mosinska, P. Marquez-Neila, M. Kozi ´ nski, and P. Fua, “Beyond ´ the Pixel-Wise
Loss for Topology-Aware Delineation,” in Conference on Computer Vision and Pattern
Recognition. IEEE, Jun. 2018, pp. 3136–3145.
67. K. Li, B. Wang, Y. Tian, and Z. Qi, “Fast and Accurate Road Crack Detection Based
on Adaptive Cost-Sensitive Loss Function,” IEEE Transactions on Cybernetics, 2021.
68. Z. Dong, J. Wang, B. Cui, D. Wang, and X. Wang, “Patch-Based Weakly Supervised
Semantic Segmentation Network for Crack Detection,” Construction and Building
Materials, vol. 258, p. 120291, 2020.
69. R. Fan, M. J. Bocus, Y. Zhu, J. Jiao, L. Wang, F. Ma, S. Cheng, and M. Liu, “Road
Crack Detection Using Deep Convolutional Neural Network and Adaptive
Thresholding,” in Intelligent Vehicles Symposium. IEEE, 2019, pp. 474–479.
70. Y. Inoue and H. Nagayoshi, “Crack Detection as a Weakly-Supervised Problem:
Towards Achieving Less Annotation-Intensive Crack Detectors,” arXiv preprint
arXiv:2011.02208, 2020.
71. S. Dorafshan, R. J. Thomas, and M. Maguire, “SDNET2018: An Annotated Image
Dataset for Non-Contact Concrete Crack Detection Using Deep Convolutional Neural
Networks,” Data in Brief, vol. 21, pp. 1664–1668, 2018.
72. S. Drouyer, “An ’All Terrain’ Crack Detector Obtained by Deep Learning on Available
Databases,” Image Processing On Line, vol. 10, pp. 105–123, 2020.
73. V. L. Guen and N. Thome, “Disentangling Physical Dynamics from Unknown Factors
for Unsupervised Video Prediction,” in Conference on Computer Vision and Pattern
Recognition. IEEE, 2020, pp. 11 474– 11 484.
74. B. Lim and S. Zohren, “Time-Series Forecasting with Deep Learning: A Survey,”
Philosophical Transactions of the Royal Society A, vol. 379, no. 2194, p. 20200209,
2021.
75. T. Sharma, P. Agrawal, N.K. Verma, Detection of dust deposition using convolutional
neural network for heritage images, in: Computational Intelligence: Theories,
Applications and Future Directions-Volume II, Springer, 2019, pp. 347–359.

356
 Review of Structural Health Monitoring Methods: A Machine Learning Approach

76. T. Masrour, I. El Hassani, M.S. Bouchama, Deep convolutional neural networks with
transfer learning for old buildings pathologies automatic detection, in: International
Conference on Advanced Intelligent Systems for Sustainable Development, Springer,
2019, pp. 204–216.
77. R. Perumal, S.B. Venkatachalam, Non-invasive detection of moss and crack in
monuments using image processing techniques, J. Ambient Intell. Human Comput.
(2020), http://dx.doi.org/10.1007/s12652-020-02006-x.
78. K. Zorlu, Description of the weathering states of building stones by fractal geometry
and fuzzy inference system in the olba ancient city (Southern Turkey), Eng. Geol. 101
(3–4) (2008) 124–133.
79. O. Akkoyun, Z. Fuat Toprak, Fuzzy-based quality classification model for natural
building stone blocks, Eng. Geol. 133–134 (2012) 66–75.

357
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

37. RFID based Smart Shopping Trolley

Mamata J. Sataraddi, Mahabaleshwar S. Kakkasageri,
Vedant Vanaki, Omkar Mutnal, Satish Bailwad,
Neelakant Vastrad
Department of Electronics and Communication Engineering,
Basaveshwar Engineering College Bagalkote, India.
Abstract:

The proposed smart shopping trolley system leverages RFID technology and IoT
connectivity to streamline the shopping experience. With an Arduino board, RFID reader
and Wi-Fi module, customers can easily navigate through stores, automatically tallying
their purchases as they go. This innovative approach minimizes wait times at checkout,
providing a seamless and efficient shopping experience. Additionally, the integration with
a centralized database and website allows for easy access to purchase history and
administrative oversight, enhancing convenience for both customers and retailers.

Keywords:

RFID, RFID Tags, Arduino, Wifi Module, Blynk app

I Introduction:

RFID is the special type wireless card which has inbuilt the embedded chip along with loop
antenna. RFID reader is the circuit which generates 125KHZ magnetic signal. This
magnetic signal is transmitted by the loop antenna connected along with circuit used to read
the RFID card number. In this work, RFID technology is utilized to streamline the shopping
experience by automating the billing process. Each product is equipped with its own RFID
card, which essentially acts as a security access card and represents the product's identity.
An RFID reader, interfaced with a microcontroller, reads these RFID cards as products are
placed in the shopping cart. The microcontroller, pre-programmed with the corresponding
card numbers, manages this process seamlessly. Additionally, a keypad is integrated with
the microcontroller to facilitate user interaction

The primary objective of the proposed work is to alleviate the inconvenience of waiting in
long queues during the billing process at shopping malls. By implementing an automatic
billing system within the shopping trolley itself, customers can significantly reduce the time
spent on checkout. Once customers finish selecting their items, the total amount is displayed
on the trolley, allowing them to conveniently pay using their pre-recharged customer card

358
 RFID based Smart Shopping Trolley

provided by the shop. Finally, all transaction information is transmitted to the central PC of
the shopping mall, ensuring efficient management of purchases and inventory. This
innovative approach not only enhances customer satisfaction but also optimizes the overall
shopping experience.

II Literature Review:

The work in [1] discusses an RFID-based Smart Shopping Cart System, aims to enhance in-
person shopping by utilizing Radio Frequency Identification (RFID) for automatic product
identification, allowing real-time viewing of total costs to reduce checkout time. The study
emphasizes the need for a technology-oriented, cost-effective, and scalable system to
improve the overall shopping experience.

The "Intelligent Shopping Cart" system, detailed in the paper [2], employs RFID and
ZigBee technologies to streamline in-person shopping. It aims to reduce time spent in malls
by automatically identifying products, updating billing in real time, and enhancing
inventory management. The system consists of three key components: Server
Communication, User Interface and Display and Automatic Billing. The integration of these
components into an embedded system offers a cost-effective and scalable solution. The
proposed model has the potential to significantly improve the shopping experience by
providing efficiency, convenience, and real-time updates.

A smart shopping system utilizing RFID technology for automated billing is presented in
[3-5]. Products with RFID tags are scanned when placed in the cart, displaying real-time
billing information. This approach minimizes manual billing, accelerates item retrieval, and
reduces waiting times. The system incorporates Raspberry Pi, Arduino, RFID tags, and a
database. Electronic Shopping Cart System utilizing RFID technology with tags on
products, an RFID reader, a microcontroller integrated with Embedded C and VB6.0
software for efficient shopping is proposed in [6].

The RFID and GSM-based Smart Trolley system is delivered in [7]. The "Smart Shopping
Trolley" using RFID technology, Zigbee modules, an ESP8266, and an LCD display is
produced in [8]. This approach enhances accuracy and speed. The work in [9] introduces a
home automation concept using ESP32 with Blynk, IR remote, and manual switches to
control 8 relays with or without internet connectivity. The system aims to improve living
standards, reduce human effort, and save energy by enabling remote control and monitoring
of home appliances via an Android-based smartphone application. Utilizing Wi-Fi
technology, the system offers accuracy, high range, and easy installation, making it user-
friendly and suitable for a wide range of applications.

With these survey it is observed that the privacy concerns may arise due to the continuous
tracking of items and user data. These technologies lead to system malfunctions or errors,
impacting the accuracy of billing. Implementation costs and maintenance may pose
financial challenges, particularly for smaller retailers. Additionally, user dependency on
smartphones and technical literacy might limit widespread adoption. To Address these
drawbacks a system smart shopping cart using RFID technology and Blynk app is proposed
in this paper for successful integration into diverse retail environments, ensuring both user
satisfaction and data.
359
International Journal of Research and Analysis in Science and Engineering

III Proposed Work:

The proposed smart shopping cart using RFID technology and Blynk app is presented in
this section. The block diagram of proposed work is as shown in figure 1.

Figure 1: Block Diagram of Proposed System

A. Components:

The details of the components used in this work are as follows:

• RFID Reader: The RFID Reader RC522 (shown in figure 2) is a popular module based
on the MFRC522 integrated circuit. It operates at 13.56 MHz and is commonly used for
reading and writing RFID tags in various projects, including access control systems,
smart locks, and inventory management solutions. The module typically communicates
with a microcontroller such as Arduino via SPI (Serial Peripheral Interface) protocol
and provides functionalities for reading RFID tags' unique identification data. It offers
a cost-effective and reliable solution for RFID-based applications.

360
 RFID based Smart Shopping Trolley

Figure 2: RFID- RC522

• RFID Tags: An RFID tag is a small electronic device comprising a microchip storing
unique identification data and an antenna for wireless communication. It enables objects
to be tracked and identified when activated by an RFID reader. These tags can be
passive, drawing power from reader signals, or active, with their power source, and
operate at various frequencies depending on the application.
• Wi-Fi Module: The ESP8266 as shown in figure 3 is a versatile Wi-Fi module widely
used in IoT (Internet of Things) projects due to its low cost and ease of use. It integrates
a microcontroller and Wi-Fi capability in a single chip, making it suitable for connecting
devices to the internet wirelessly. The module supports various communication
protocols and can be programmed using Arduino IDE or other development
environments. With its small form factor and low power consumption, the ESP8266
enables devices to communicate and exchange data over Wi-Fi networks, facilitating
remote monitoring, control, and data transfer in IoT applications.

Figure 3: Wi-Fi Module – ESP8266

• Arduino: The Arduino Uno shown in figure 4 is a popular microcontroller board widely
used in electronics projects. It features an Atmega328P microcontroller, digital and
analog input/output pins, a USB interface for programming and communication, and a
power jack for an external power supply. The Uno is compatible with a wide range of
sensors, actuators, and other components, making it ideal for prototyping and DIY
projects.
361
International Journal of Research and Analysis in Science and Engineering

Figure 4: Arduino Uno

• LCD Display: An LCD (Liquid Crystal Display) display is a flat panel (shown in figure
5) that uses liquid crystals to produce images or text. It's commonly used in electronic
devices like digital clocks, calculators, and consumer electronics for showing
information. In projects, an LCD can be connected to a microcontroller like Arduino to
provide visual feedback or display data in real time.

Figure 5: LCD Display

• IR Sensor: An IR (Infrared) sensor is a device that detects infrared radiation emitted

by objects. It's commonly used in various applications such as motion detection,
proximity sensing, and object detection. In projects, an IR sensor can be used to detect
motion or presence, allowing for the automation of tasks or the creation of interactive
systems.

B. Flow Chart:

The flow diagram of the proposed model is as shown in figure 6.

362
 RFID based Smart Shopping Trolley

Figure 6: Flow Chart of Proposed Model

C. Software Used:

Blynk is a platform that enables developers to create IoT (Internet of Things) applications
easily. It provides a framework for building apps for controlling hardware remotely through
smartphones or tablets. With Blynk, a user interface is created to interact with your IoT
devices, such as Arduino, Raspberry Pi, ESP8266 and others, shown in figure 7 without
much coding. It simplifies the process of connecting hardware to the internet and creating
mobile apps to control them. The proposed Smart Shopping Cart, utilize Blynk to create a
user-friendly interface for controlling and monitoring the cart remotely. Blynk provides a
mobile control interface, allowing users to interact with the Smart Shopping Cart using their
smartphones. With Blynk, users can monitor the items in their shopping cart in real-time,
including scanned products, total bill, and payment status. Blynk widgets such as buttons
and sliders enable users to add or remove products from the cart remotely, triggering actions
such as RFID scanning and bill updates. This work leverage Blynk's customizable user
interface to create a tailored experience for our users, incorporating various widgets like
buttons, displays, and indicators. Blynk's cross-platform compatibility ensures that users
can control the Smart Shopping Cart using their preferred mobile platform, whether iOS or
Android. Additionally, Blynk's cloud connectivity allows for remote control of the shopping
cart from anywhere in the world, as long as there is an internet connection. With Blynk's
secure data transmission protocols, we ensure that all communication between the mobile
app and the Smart Shopping Cart is encrypted, protecting user privacy and sensitive
information. Overall, Blynk simplifies the development of the software end of our Smart
Shopping Cart project, providing a seamless and intuitive user experience for our customers.
363
International Journal of Research and Analysis in Science and Engineering

Figure 7: Hardware Connection Diagram with Blynk App

III Results:

The screenshots of the results obtained by the work are shown below.

364
 RFID based Smart Shopping Trolley

365
International Journal of Research and Analysis in Science and Engineering

IV Conclusion:

The Smart Shopping Cart project represents a significant step forward in modernizing the
retail experience. By integrating RFID technology, Blynk software, and a user-friendly
interface, a system has developed that simplifies and enhances the shopping process for both
customers and retailers. This innovative solution addresses several key challenges faced by
the retail industry, including long checkout queues, manual inventory management, and
inefficiencies in the shopping process.

With real-time monitoring, remote control capabilities, and seamless integration with
existing retail infrastructure, the Smart Shopping Cart project offers numerous benefits.
Customers benefit from a streamlined and efficient shopping experience, with automatic
product scanning, real-time bill updates, and convenient payment options. Retailers benefit
from increased operational efficiency, reduced waiting times, and improved inventory
management, leading to cost savings and enhanced customer satisfaction.

By continuing to innovate and improve the Smart Shopping Cart project, it can be aimed to
revolutionize the retail experience and set new standards for efficiency, convenience, and
customer satisfaction in the industry.

References:

1. Shefali Gupta, “Arduino Based Smart Cart,” International Journal of Advanced

Research in Computer Engineering and Technology (IJARCET), ISSN:2278-1323,
Volume 2, Issue 12, pp. 3083-3090, December 2013
2. Sainath S, Automated Shopping Trolley for Super Market Billing System, International
Conference on Communication, Computing and Information Technology, 2014
3. Vinoth Kumar, Poornima, “Smart Shopping Trolley”, International Journal of
Advanced Research in Computer and Communication Engineering, May 2017
4. HenryOhize, David Michael “A new automated smart cart system for modern shopping
centres”, Bulletin of Electrical Engineering and Informatics, Vol.10, No. 4, pp-2028-
2036, 2021
5. Mrs. R. Hemalatha, A. Krithika, “RFID and GSM Based Smart Trolley”, International
Journal of Advanced Research Trends in Engineering and Technology (IJARTET), Vol.
3, Issue 3, March 2017

366
 RFID based Smart Shopping Trolley

6. Chaithra, Ankita, Aishwarya” BillSmart - A Smart Billing System using Raspberry Pi

and RFID”, International Journal of Advanced Research in Computer Engineering and
Technology (IJARCET), Dec-2021
7. Kalyani Dawkhar, Shraddha Dhomase, Samruddhi Mahabaleshwarkar, “Electronic
Shopping Cart for Effective Shopping based on RFID”, International Journal of
Advanced Research Trends in Engineering and Technology (IJARTET), Jan-2018
8. Mr.P. Chandrasekar and Ms. T. Sangeetha “Smart Shopping Cart with Automatic
Billing System through RFID and ZigBee”, IEEE International Conference on
Information Communication and Embedded Systems (ICICES2014), PP. 1-4, Chennai,
2014
9. Mamata Khatu, Neetu Kaimal, Pratik Jadhav, Syedali Adnan Rizvi, “Implementation
of Internet of Things for Home Automation”, International Journal of Technology,
Volume 3, Issue 2, February 2015.

367
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

38. Smart Home Automation using IoT

Anjali Honakeri, Niveditha M., Sweta M. Elangadi,
Priyadarshini Jalikatti, Chayalakshmi C. L.
Department of Electrical and Electronics Engineering,
Basaveshwar Engineering College,
Bagalkote, Karnataka, India.

ABSTRACT:

The rapid advancement of Internet of Things (IoT) technology has revolutionized the
concept of home automation, transforming traditional homes into intelligent, connected
ecosystems. This paper presents an in-depth exploration of smart home automation utilizing
IoT, focusing on the design, implementation, and potential applications. The system
integrates various IoT devices, sensors, and actuators to monitor and control home
environments efficiently. The effectiveness of the proposed smart home automation system
is validated through experimental setups and real-world scenarios. Results demonstrate
significant improvements in convenience, energy efficiency, and security compared to
conventional home automation systems. This research provides a comprehensive
framework for future developments in IoT-based smart home technologies, paving the way
for more intelligent and responsive living environments.

KEYWORDS:

Smart Home Automation, Internet of Things (IoT), Sensors, Actuators, Energy Efficiency,
Security, Privacy, Cloud Services, Remote Monitoring.

I Introduction:

IoT is an environment made up of physically connected, network-connected items that have

been given IP addresses and can connect to networks without the need for human
interaction. It can transmit data across a network without requiring communication between
people or between people and computers.

A collection of hardware components, such as microprocessors, sensors, and other

components, make up an Internet of Things system. These components communicate data
to and from the microcontroller and server. Real-time monitoring and control of home
equipment can be accessed and managed by users using a mobile or online application.

368
 Smart Home Automation using IoT

The concept of home automation has evolved significantly over the past few decades.
Initially, it involved the simple remote control of home devices such as lighting, heating,
and appliances. However, the advent of the Internet of Things (IoT) has catalyzed a
paradigm shift, transforming traditional home automation into sophisticated smart home
systems. IoT enables devices to communicate with each other and with centralized control
systems via the internet, creating interconnected networks that enhance the functionality,
efficiency, and convenience of home environments.

The journey of home automation began with simple mechanical devices and manually
controlled systems. Over time, these evolved into electrically operated devices with limited
remote control capabilities. The true revolution came with the advent of IoT, which
introduced the ability for devices to communicate over the internet, share data, and be
controlled remotely. This has led to the creation of highly sophisticated smart home systems
that offer unparalleled convenience, efficiency, and security.

Systems that are hardwired or wireless can be used to set up smart houses. Smart home
technology offers financial savings and convenience to homeowners. Producers and
consumers of smart home equipment are still beset by security vulnerabilities and glitches.
Homeowners can begin using smart home devices with smaller individual gadgets that cost
less than $100, even if full-scale home automation may cost thousands of dollars. This paper
explains the monitoring and controlling of device settings using communication modules,
sensors, actuators, and microcontrollers.

II Literature Review:

Several technical papers are reviewed before taking up this work to identify the gaps. This
section of the paper explains the literature survey.

The paper presents an advanced IoT-based home automation system that leverages Google
Assistant and the Thing Speak IoT platform. The system aims to enhance the convenience,
security, and energy efficiency of modern homes by integrating voice control and real-time
data processing capabilities.

This paper provides a comprehensive framework for developing advanced IoT-based home
automation systems, paving the way for future innovations in smart home technologies. The
paper concludes that integrating Google Assistant with the Thing Speak IoT platform offers
a robust and user-friendly solution for advanced home automation. The system's ability to
process voice commands, analyze real-time data, and automate responses significantly
enhances the smart home experience. Future work includes expanding the system's
capabilities and exploring additional applications in smart home technology [1].

This paper presents the design and implementation of an interactive Internet of Things
(IoT)-based home automation system controlled by speech. This paper provides a detailed
framework for developing speech-controlled home automation systems, highlighting the
potential for future advancements in making home environments more accessible and
convenient through IoT and voice technology integration. The system aims to improve the
convenience and accessibility of home automation by enabling users to control home
appliances through voice commands.
369
International Journal of Research and Analysis in Science and Engineering

The system utilizes speech recognition technology to interpret voice commands from users.
It converts spoken words into digital commands that can be processed by the system. The
system integrates various home appliances into a centralized system via an IoT network and
allows remote control and monitoring of devices through the internet. The system
architecture includes modules for speech processing, data acquisition, data processing, and
device control. Speech commands are captured using a microphone and processed using a
speech recognition module (e.g., Google Speech API). Recognized commands are
transmitted to a microcontroller, which then controls the respective home appliances via
relays or other interfacing circuits.

The system incorporates security measures such as encrypted communication channels and
user authentication to protect against unauthorized access and data breaches. Experimental
results demonstrate the system's effectiveness in accurately recognizing and executing voice
commands. User feedback indicates a high level of satisfaction with the convenience and
functionality provided by the speech-controlled system. The paper discussed about the
interactive IoT-based speech-controlled home automation system significantly enhances
user convenience and accessibility. The integration of speech recognition with IoT
technology offers a robust solution for modern home automation needs. Future work
includes expanding the system’s capabilities, improving speech recognition accuracy, and
exploring additional applications for the technology [2].

The paper discusses the development of a smart home automation system leveraging the
Internet of Things (IoT). This paper offers a detailed framework for developing smart home
automation systems using IoT, highlighting the potential for enhancing user experience and
energy efficiency through advanced technology integration. The system is designed to
provide users with enhanced control, convenience, security, and energy efficiency in
managing home appliances and systems.

The system integrates various home appliances and sensors into a cohesive network and
facilitates remote control and monitoring via internet connectivity. The system is composed
of interconnected modules for data acquisition, processing, and control. Devices are
connected to microcontrollers which relay data to a central server. The server processes data
and issues control commands based on user inputs and predefined rules.

The system allows users to control lighting, heating, air conditioning, security cameras, and
other appliances remotely. The system implements automation rules (e.g., turning off lights
when no motion is detected) to enhance energy efficiency. A mobile app is developed to
provide an intuitive interface for managing home automation which supports real-time
notifications and remote control capabilities. The system optimizes energy usage by
automating control of appliances based on sensor data and reduces energy consumption
through intelligent scheduling and real-time adjustments. Experimental results validate the
system's functionality in terms of reliability, responsiveness, and user satisfaction.
Performance metrics indicate efficient communication and accurate execution of control
commands [3].

The paper explores the development of an IoT-based home automation system aimed at
improving the convenience, security, and efficiency of managing household appliances. The
proposed system leverages IoT technology to provide remote monitoring and control

370
 Smart Home Automation using IoT

capabilities, making home automation more accessible and effective. This paper presents a
comprehensive approach to developing an IoT-based home automation system,
demonstrating the potential to enhance user experience and operational efficiency through
advanced technology integration. The system integrates various household devices into a
unified network and also enables remote access and control through internet connectivity.

The system consists of interconnected modules for data collection, processing, and control.
Devices are connected to microcontrollers, which send data to a central server for
processing. The server issues control commands based on user inputs and predefined
automation rules. The system allows users to control lighting, HVAC systems, security
cameras, and other appliances remotely.

The system developed supports automation scenarios like turning off lights when no motion
is detected or adjusting the thermostat based on ambient temperature. The user interface
provides easy access to control and monitor home devices. Features like real-time
notifications and control options for enhanced user convenience are also provided. The
automation system optimizes energy use by adjusting device operations based on real-time
sensor data. It reduces energy consumption through intelligent automation and user-defined
schedules. Experimental results show the system's effectiveness in providing reliable and
responsive control over home appliances. User feedback indicates high satisfaction with the
system's functionality and ease of use [4].

But these literatures concentrated on a single goal of using IoT for home automation. We
thought of developing a system with the following objectives along with using IoT without
compromising the comforts provided to the users:

• IoT based home automation for monitoring resistive and inductive loads. The lights and
fans in a room are controlled remotely by the mobile app
• Automatic turns on and off of the lights and fans in a room based on the presence or
absence of a person in a room
• Automatic speed control of a fan based on the temperature of the room

III Methodology:

The proposed system uses sensors such as PIR motion sensor and DHT11 temperature
sensor to gather the information about the environment in the room. The developed mobile
app helps the user to know about the status of the lights and fans in a room remotely and
provides the facility to control them also. The block diagram of the proposed system is
shown in Figure 1.

The system consists of Arduino Uno, Wi-Fi, sensors, relays and resistive and inductive
loads such as bulbs and fans. The mobile app is created from the software studio and the
number of bulbs and fans are incorporated. The mobile app is designed to provide users
with an intuitive and efficient way to control lights and fans in their home remotely.
Leveraging IoT technology, the app connects to smart devices installed in the home,
allowing for real-time monitoring and control.

371
International Journal of Research and Analysis in Science and Engineering

Figure 1: Block Diagram of the System

Load Controlling:

• Create software from the studio and build the front-end mobile app.
• Dump the program in Arduino IDE which connects the Arduino and mobile through
Wi-Fi connection.
• After WIFI connection is established between mobile and Arduino, the Arduino is able
to receive the command from the mobile app
• The relay connected to Arduino is capable of controlling the resistive and inductive
loads.
• The loads are monitored using the mobile app by switching the on/off condition.

Temperature Based Fan Speed Control:

• The temperature sensor (such as DHT11) connected to Arduino will measure the
temperature.
• A fan is connected to a PWM capable pin on the Arduino is to control its speed.
• A program in Arduino IDE is written to read the temperature and humidity from the
sensor and control the fan speed accordingly.
• The program will determine the desired fan speed based on the temperature reading.
• User has to define temperature thresholds for different fan speeds (low, medium, high).
• The system will increase fan speed if the temperature exceeds a threshold.
• The system developed will decrease fan speed if the temperature falls below a threshold.
• The system will turn off the fan if the temperature is below a certain minimum
threshold.

Motion Sensor:

• PIR sensor is used to detect whether a human has moved in or out of sensor range.
• PIRs are basically made of a pyroelectric sensor which detect the levels of infrared
radiation.
372
 Smart Home Automation using IoT

• The sensor detects the motion by comparing the changes in infrared radiation.
• The sensor amplifies and processes the detected signal.
• When sensor detects motion, it triggers an output signal. This output can be used to
activate a relay, turn on the light and send a signal to a microcontroller or other devices.
• The PIR sensor is connected to the power supply
• The system is tested for various lighting conditions to ensure reliable operation.
• It is necessary to regularly monitor the sensor’s performance and conduct maintenance
as needed to ensure continued operation.

Flowchart:

The flowchart begins with creating the application in Android Studio for the mobile's front
end and then proceeds to select the WIFI module (the Arduino UNO), as shown in Figure
2. If a WIFI connection is established between the mobile app and the Arduino UNO,
continue to the next step to provide the input command to the relay; otherwise, return to
step 2. Upon receiving the command from the Arduino UNO to the relay, you can control
home appliances through the mobile app.

Figure 2: Flow chart

373
International Journal of Research and Analysis in Science and Engineering

In the first condition, if the mobile app input is on, the bulb turns on; otherwise, it turns off.
In the second condition, when a motion-based sensor detects an object, it triggers the output
and sends a command to turn on the light. In the third condition, the DHT11 connected to
the Arduino (UNO) reads the temperature. If the temperature exceeds the threshold, the fan
runs at high speed; if it's below the threshold, the fan runs at low speed.

Figure 3: Working Model of the System

IV Results and Discussions:

The proposed work is successfully completed and following objects are achieved.

• The Resistive load is successfully controlled through the developed mobile app.
• The motion-based sensor (Passive infrared sensor) detects the human being present in
front of sensor.
• The DHT11 sensor detects the temperature of the room and controls the speed of fan.

Figure 4: Developed Mobile App

374
 Smart Home Automation using IoT

V Conclusion:

The exploration and implementation of home automation systems using IoT technology
demonstrate significant advancements in enhancing convenience, security, and energy
efficiency in modern households. By integrating various household appliances into a
cohesive IoT network, users gain unprecedented control and monitoring capabilities,
accessible from anywhere at any time. The ability to control lights, fans, and other
appliances remotely through a mobile application or voice commands simplifies daily
routines. Automation features such as scheduling and scene creation further streamline
household management, allows users to customize their environment effortlessly. IoT-
based home automation systems incorporate robust security measures, and real-time alerts.
Intelligent automation and real-time monitoring enable significant energy savings by
optimizing the operation of home appliances. Features like energy usage tracking and
automated power management contribute to more sustainable living practices.

References:

1. Reddy, N. Negi, Z. Gupta, S. Sood, I. Kansal, N. Aggarwal, “Advanced IoT Home

Automation using Google Assistant and Thing Speak IoT Platform," 2nd International
Conference on Advance Computing and Innovative Technologies in Engineering
(ICACITE), Greater Noida, India, 2022, pp.426-432,
Doi:10.1109/ICACITE53722.2022.9823620.
2. N. C. Noruwana, P. A. Owolawi, T. Mapayi, “Interactive IoT-based Speech-Controlled
Home Automation System,” 2nd International Multidisciplinary Information
Technology and Engineering Conference (IMITEC), Kimberley, South Africa, 2020,
pp. 1-8, doi:10.1109/IMITEC50163.2020.93340.
3. U. Singh and M. A. Ansari, “Smart Home Automation System Using Internet of
Things,” 2nd International Conference on Power Energy, Environment and Intelligent
Control (PEEIC), Greater Noida, India, 2019, pp.144149,
doi:10.1109/PEEIC47157.2019.8976842.
4. H. K. Singh, S. Verma, S. Pal, K. Pandey, “A step towards Home Automation using
IoT,” Twelfth International Conference on Contemporary Computing (IC3), Noida,
India, 2019, pp. 1-5, Doi: 10.1109/IC3.2019.8844945.
5. M. Al-Kuwari, A. Ramadan, Y. Ismael, L. Al- Sughair, A. Gastli, M. Benammar,
“Smart- Home Automation using IoT-Based Sensing and Monitoring Platform," IEEE
12th International Conference on Compatibility, Power Electronics and Power
Engineering (CPE- POWERENG 2018), Doha, Qatar, 2018, pp. 1-6,
doi: 10.1109/CPE.2018.8372548
6. Chan, M., Campo, E., Esteve, D., Fourniols, J.Y., “Smart homes-current features and
future perspectives,” Maturitas, Vol. 64, Issue 2, 2009, pp. 90- 97.

375
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

39. Speech Intelligibility Enhancement based on

Spectral Splitting Technique
Aparna Chilakawad
Electronics and Communication Engineering Department,
Basaveshwar Engineering College, Bagalkot.
Pandurangarao N. Kulkarni
Visvesvaraya Technological University,
Belagavi Karnataka, India.
Abstract:

People with moderate sensorineural hearing loss (MSNHL) have difficulty in recognizing
speech in loud environment because of masking. There is no medical treatment for this loss.
By spectrally dividing the voice signal using a pair of time varying FIR comb filters (TVCF)
for binaural hearing aid, the effect of frequency masking can be reduced. As a result,
perception of voice signal is increased. Using a frequency sampling method TVCF of order
512 are designed. These filters are having 22 octave bands (one-third) varying from
frequency 0 kHz to 11 kHz. Magnitude responses of these filters are complementary to one
another which sweep along the frequency axis with time shift less than just noticeable
difference (JND). This enhances gap identification capability while maintaining the
advantages of the frequency splitting method. To evaluate effectiveness of frequency
splitting scheme, the Modified Rhyme Test (MRT) is used for speech intelligibility test. This
test is examined on bilateral MSNHL subjects. Three hundred monosyllabic CVC
(Consonant Vowel Consonant) syllables, serve as the test signals, are utilized to evaluate
the scheme. The findings show that in environments with higher levels of noise, people are
better able to comprehend and interpret processed speech.

Keywords:

Masking, Time-varying FIR filter, Spectral splitting, MRT.

I Introduction:

Sensorineural deafness occurs because of injury to the hair cells of cochlea in the inner ear
or the nerve that connects the inner ear to the brain. Frequency masking is one of the
characteristics of sensorineural hearing loss (SNHL) [1]. Many researchers showed that this
masking can be reduced by using a pair of FIR comb filters with complementary magnitude
responses to split the speech spectrum into two parts with complementary spectral
components for binaural hearing aids [2-3].
376
 Speech Intelligibility Enhancement based on Spectral Splitting Technique

The scheme enhanced the ability to perceive speech. The aim of works [2-3] was to improve
speech perception in moderate sensorineural hearing loss (MSNHL) persons using spectral
splitting scheme. In the work [2] filters were designed with a sampling frequency of 10 kHz
with 513 filter coefficients using frequency sampling method. The methodology was
assessed on normal hearing subjects as well as bilateral MSNHL subjects. For binaural
hearing aid, the scheme was useful.

The approximate 4:2 compressor adders in the memory-less DA-based FIR filter
architecture were suggested in the article for hearing aid [4]. In DA (Distributed Arithmetic)
architecture, increasing the filter order caused the ROM's size to grow gradually.
Compressor adders were used in the design of memory-less DA as a way to reduce power
consumption and area of the FIR filters, creating the necessary area and power reduction
for the use of hearing aids. The suggested DA-based FIR filter architecture was created
utilizing the Synapsis Application Specific Integrated Circuit Design Compiler on 90 nm
technology. Comparing the suggested design to systolic architecture, there was a 45%
decrease in area delay product, and compared to OBC (Offset Binary Coding) DA
architecture, there was a 10% reduction in ADP. The results indicated that the suggested
architecture has fewer slices than the best current solutions. The suggested design also
incorporates a FPGA. Using MATLAB Simulink, the suggested design was applied as a
decimation filter to eliminate undesired signals in hearing aid applications.

For the purpose of listening, the study [5] suggested a hearing loss system model with a
variable bandwidth FIR filter (VBF) and adaptive algorithms. The purpose of the adjustable
band filter was to deliver the right amount of sound. There were several sub-filters in this
filter, and each one was developed with a specific set of bandwidths. By trial and error, the
sub-bands were acquired and their magnitudes were adjusted appropriately. To enhance the
signal's quality, algorithms like Recursive Least Squares (RLS), Normalized Least Mean
Squares (NLMS), and Least Mean Squares (LMS) were used. The multiple bandwidth
filters were analyzed with different degree of loss. The difference between the ideal and
actual responses was computed to find the matching error. The results demonstrated that the
suggested filter offered a minimal matching error and was between 0 and 2.5 dB.

The study [6] aims to investigate the combined effects of amplitude compression, frequency
splitting, and adaptive Wiener filter on the intelligibility of speech for individuals with
hearing impairments in unfavorable listening conditions. The adaptive Wiener filter
operated by varying the filter frequency response between samples in accordance with the
statistics of the speech signal (mean and variance). To reduce the impact of spectral
masking, the speech signal was spectrally separated using the filter bank technique.
Amplitude compression was used with a compression factor of 0.6 in order to solve the
issue of hearing loss. The intelligibly measurement was done with MRT (Modified Rhyme
Test) measure on MSNHL people. There were three hundred consonant-vowel-consonant
(CVC) words in the test protocol. The results indicated that processing with this scheme
enhanced speech intelligibility when compared to an adaptive Wiener filter. A maximum
improvement in speech recognition score of 32.935 percent was noted at a lower SNR value
of -6 dB. Therefore, in loud circumstances, the combined method helped those with hearing
impairments to understand speech better.

To make speech more understandable for near-end listeners in noisy environments, a speech
pre-processing technique was described [7]. Based on a spectro-temporal auditory model,

377
International Journal of Research and Analysis in Science and Engineering

the method increased the intelligibility by redistribution of speech energy over time and
frequency for a perceptual distortion measure.

The two objective predictors of intelligibility were used for both before and after processing.
They were coherence speech intelligibility index (CSII) and short-time objective
intelligibility (STOI). To judge the effectiveness of the suggested approach voice quality
was reduced using F-16, babble, factory and white noise within a -15 to 5 dB SNR range.
CSII score for unprocessed speech (with white noise) was 0.4 and for processed speech was
0.65. Results showed that there was an improvement in intelligibility with the proposed
concept.

The work [8] introduced an effective method that improved clean voice intelligibility in
noisy environment. In order to restore voice intelligibility, the algorithm increased the
average speech spectrum over the average noise spectrum. It was taken care in the algorithm
to prevent the hearing damage from increased speech spectrum. The Speech Intelligibility
Index (SII) was used to assess the performance of suggested algorithm. In addition to white
noise and destroyer engine noise from the NOISEX-92 database, the SII was computed for
each speech file in the TIMIT database. A time adaptive and frequency dependent SNR
recovery approach was presented. The SII was same for SNR 15dB to 20dB or increased by
0.5. This indicated improved speech intelligibility.

For individuals with sensorineural hearing impairment, a multiband frequency compression

(MFC) approach in addition to noise reduction was developed to neutralize the spectral
masking effect [9]. Noise reduction technique i.e. Wiener filter and spectral subtraction
approaches were used in combine technique. MRT was used to measure the intelligibility
of processed speech in the presence of additive noise and tested on both normal hearing and
hearing-impaired individuals. As compared to spectral subtraction technique with an MFC
scheme, the results of MRT processing for the compression factor of 0.6 using a cascading
wiener filter with an MFC scheme on hearing-impaired subjects showed a maximum
improvement in speech intelligibility of 25.92% to 30.134 % and a decrease in response
times of 0.815 to 1.626 seconds for SNR values of + 6 dB to - 6 dB in steps of + 3 dB,
respectively.

Since sensorineural hearing loss cannot be treated medically, the purpose of the current
study is to divide the speech spectrum using a pair of time-varying FIR comb filters with
complementary magnitude responses to enhance speech perception in MSNHL people. A
time-varying filter's magnitude response (Hm(f)) and impulse response (Hm(n) are not
static, making them function of time 'm'. We suggested a continuous shift in magnitude
responses with time shifts chosen below just noticeable difference (JND) for binaural
dichotic presentation. Using this filter feature, additionally, gap identification capability in
speech is enhanced without negating the benefit of the frequency splitting technique. Thus,
speech perception is enhanced.

II Designing:

The processing of signal block diagram is shown in figure 1. Using a pair of TVCF of order
512 that have complementary magnitude responses to one another, a voice signal is
separated into two signals with complementary spectral components to present idiotically.
MATLAB programming software is used for developing filters. Frequency sampling
378
 Speech Intelligibility Enhancement based on Spectral Splitting Technique

approach [10] of filter design is used with a sampling frequency of 22 kHz. The following
is the FIR filter's system function which is static with respect to time.

2π𝑓
|𝐻(𝑓)| = √2(1 − 𝑐𝑜𝑠 ( )
𝑓𝑠

The dynamic magnitude response of the time-varying FIR filter is represented by |Hm(f)|.
There are 11 pass bands and 11 stop bands in time-varying FIR filters, and are based on
octave bands (one-third). A sampling frequency of 22 kHz is considered to have better
speech quality. Larger bands at higher frequencies and narrow bands at lower frequencies
can be seen in 1/3-octave bands. For higher frequency bands, the transition bandwidth is
varied between 70 and 80 Hz. Figure 2 shows the superimposed complementary magnitude
responses of a pair of time-varying FIR comb filters at various points of time. The time shift
of below JND is taken into consideration for the continuous shift in magnitude responses.
Crossover gain is kept between -6 dB and -6.5 dB for bands with higher frequencies and
between -4 dB and -7 dB for bands with lower frequencies in order to minimize the variation
in perceived loudness.

Figure 1: Signal processing with a pair of time-varying FIR comb filters.

379
International Journal of Research and Analysis in Science and Engineering

(a)

(b)

(c)

Figure 2 (a), (b) and (c) superimposed complementary magnitude responses of a pair
of time-varying FIR comb filters at various points of time.

III Results:

The method of spectral splitting of speech signal employing a pair of time-varying comb
filters with the continuous sweep in magnitude responses with a time shift below JND is
tested for speech intelligibility using an MRT test on six bilateral MSNHL listeners utilizing
300 monosyllabic CVC words as test signals. The guy uttered the CVC words, in an
audiometry room and words were recorded using a B&K microphone model No. 2210 with
a sampling frequency of 10 KHz and 16-bit quantization. "Would you write…." heard
before every CVC word. Six bilateral MSNHL subjects between the ages of 30 and 49(three
380
 Speech Intelligibility Enhancement based on Spectral Splitting Technique

men and three women) took part in hearing tests to evaluate intelligibility. Bilateral MSNHL
audiometric level is shown in table1. The testing procedure was carried out using an
automated test administration system. The subjects received preliminary information on the
testing protocol and stimuli. Once subjects were comfortable with the testing process, CVC
words were presented to them. These words were presented at a reasonable speech signal
volume.

A total of 1800 words (300 words x 6 SNR levels) were presented to each participant to
respond. The voice recognition score (in percentage terms) and response time (in seconds)
were recorded using a random file. All 300 words were grouped into six test lists, each
containing 50 words. Each test list's words were chosen using a two-level randomization
procedure to overcome biasing (i.e.1x, 1y, 2x, 2y, 3x, 3y; here, x, y indicates the word level
inside each set while the number indicates the set level). The test was conducted for about
a month, depending on the subject's interest and availability. Table 2 presents listening test
results for SNHL individuals with MRT materials. The average speech recognition score
(%) for both processed and unprocessed speech at different SNR conditions is shown.
Comparing the average recognition scores of processed speech to the unprocessed speech
average recognition scores at SNR levels (∞ dB, 6 dB to -6 dB in step of 3dB), the
improvements in processed speech recognition scores are -0.44,17.06, 22.44, 24.83, 26.11
and 28.61 in % age respectively.

The average response time (in second) of the same participants for the given SNR levels are
shown in table 3, where it is evident that processed speech response time is less compared
to unprocessed. The reductions in response time are -0.049, 0.603, 0.904, 0.950, 1.083 and
1.125 in seconds respectively at defined SNR.

Table 1: Average Threshold Level for The Subjects with SNHL (Bilateral Moderate)
In Audiogram

Subjects Sex & Age Ear Hearing threshold level ( dB HL)

Frequency
250 Hz 500 Hz 1000 Hz 2000 Hz 4000 Hz 8000 Hz
1
S1[M,40] R 60 45 55 40 60 65
2
L 40 55 55 50 60 60
S2[M,49] R 50 55 50 45 55 60
L 55 50 45 55 50 55
S3[M,41] R 20 30 60 40 50 50
L 20 30 60 50 40 50
S4[F,33] R 65 60 55 50 40 45
L 60 55 55 50 45 50
S5[F,30] R 25 30 40 50 60 90
L 20 30 40 50 60 85
S6[F,38] R 35 45 55 60 65 70
L 25 40 45 50 60 60
381
International Journal of Research and Analysis in Science and Engineering

1
R: Right ear, 2L: Left ear

Table 2: Speech Recognition Scores (In Percentage) For Bilateral SNHL Subjects

Subjects SNR (dB)

Sex & Age
∞ dB 6 dB 3 dB 0 dB -3 dB -6 dB
3 4
Unpr Pro Unpr Pro Unpr Pro Unpr Pro Unpr Pro Unpr Pro
S1[M,40] 88 86.67 67 83.00 59.33 79.67 50.33 76.00 44.66 71.33 41 71.33
S2[M,49] 86 85.67 67.66 85.67 59.66 81.00 51 76.00 42.66 70.00 45 73.67
S3[M,41] 87 86.00 67.33 84.33 60 81.67 49.33 72.00 45 69.00 41.66 69.33
S4[F,33] 86 87.00 66.33 83.00 61 83.00 50.66 73.33 46.33 71.00 41.33 68.00
S5[F,30] 85.66 85.67 66 83.00 57 82.67 48.66 75.00 45 72.00 40 68.00
S6[F,38] 86.33 85.33 67.33 85.00 59.33 83.00 49.66 76.33 43.33 70.33 41.33 71.67
Avg. 86.5 86.06 66.94 84.00 59.38 81.83 49.94 74.78 44.49 70.61 41.72 70.33
Improvement -0.44 17.06 22.44 24.83 26.11 28.61

3
Unpr: Unprocessed, 4Pro: Processed

Table 3: Response Time (In Seconds) For Bilateral SNHL Subjects.

5
Sub. Response Time in seconds
SNR in dB
∞ +6 +3 0 -3 -6
Unpr Pro Unpr Pro Unpr Pro Unpr Pro Unpr Pro Unpr Pro
S1 3.459 3.168 4.778 3.423 4.413 3.154 4.635 3.329 4.768 3.687 5.339 4.532
S2 3.758 3.855 3.874 3.437 4.394 3.653 3.613 3.169 4.838 3.421 5.135 4.452
S3 3.658 3.633 4.774 3.782 3.484 3.132 4.945 4.243 4.484 3.487 5.873 4.487
S4 3.474 3.321 3.749 3.523 4.937 3.145 4.957 3.623 4.887 3.839 4.964 3.682
S5 4.341 4.629 3.563 3.389 4.335 3.739 4.434 3.221 4.686 3.483 4.733 3.234
S6 3.857 4.237 3.874 3.442 4.334 3.653 3.914 3.216 4.347 3.598 5.481 4.387
Avg. 3.757 3.807 4.102 3.499 4.316 3.413 4.416 3.467 4.668 3.586 5.254 4.129
6
Redn. -0.049 0.603 0.904 0.950 1.083 1.125

5
Sub: Subjects, 6Redn: Reduction

IV Conclusion:

A pair of time-varying comb filters (FIR) can be used to divide a speech signal into two
complementary signals for binaural dichotic presentation. The magnitude responses of these
filters are continuously change along frequency axis with time shift below JND, helping to
reduce the frequency masking effect and enhance speech perception. The suggested method
is evaluated for intelligibility using the MRT test conducted on bilateral SNHL subjects.

382
 Speech Intelligibility Enhancement based on Spectral Splitting Technique

The outcome demonstrated a 28.61% increase in intelligibility and 1.125 seconds decrease
in response time at SNR -6dB. The proposed approach with a pair of time-varying comb
filters is found to work better in a noisy environment.

References:

1. Brian C. J. Moore, “An Introduction to Psychology of Hearing”, 4th ed. London:

Academic,1997.
2. P. N. Kulkarni, et al.” Binaural dichotic presentation to reduce the effects of spectral
masking in moderate bilateral sensorineural hearing los”, International Journal of
Audiology, pp. 334-344, Vol 51, No. 4, 2012 doi: 10.3109/14992027.2011.642012.
3. Cheeran, Alice N.et al., "Design of comb filters based on auditory filter bandwidths for
binaural dichotic presentation for persons with sensorineural hearing impairment."
International Conference on Digital Signal Processing Proceedings. DSP 2002 (Cat.
No. 02TH8628), pp. 971-974. Vol. 2 IEEE, 2002.
4. Rammohan, S. et al., "High performance hardware design of compressor adder in DA
based FIR filters for hearing aids." International Journal of Speech Technology pp.807-
814,2020
5. Mary, Sherin. "Multiple Bandwidth FIR Filter Design with Adaptive Algorithms for
Hearing Aid Systems." Indian Journal of Pure & Applied Physics (IJPAP), pp 605-623,
Vol 58, No. 8, 2020
6. Rajani Pujar, "Combined Effect of Adaptive Wiener Filter and Spectral Splitting of
Speech Signal in Improving Speech Intelligibility for Hearing Impaired
People," International Journal of Future Computer and Communication pp. 61-66,
Vol. 11, No. 3,2022.
7. Taal, et al., "A speech preprocessing strategy for intelligibility improvement in noise
based on a perceptual distortion measure."2012 IEEE International Conference on
Acoustics, Speech and Signal Processing (ICASSP), pp. 4061-4064. IEEE, 2012.
8. Sauert, Bastian, and Peter Vary. "Near end listening enhancement: Speech intelligibility
improvement in noisy environments."2006 IEEE International Conference on
Acoustics Speech and Signal Processing Proceedings, pp. I-I. Vol. 1, IEEE, 2006.
9. Pujar, Rajani S., P. N. Kulkarni. "Speech Intelligibility Improvement based on Noise
Reduction and Frequency Compression Technique." 2021 IEEE 18th India Council
International Conference (INDICON), pp. 1-6. IEEE, 2021.
10. J.G. Proakis, D.G. Manolakis,” Digital Signal Processing Principles, Algorithm and
Applications”. New Delhi: Prentice Hall,1997.

383
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

40. Survey on Health Monitoring System

Kartik Kulkarni, Shrinidhi Joshi,
Chandrashekhar G. Hadalagi,
Shivananda, Anand H. Unnibhavi
Dept. of Electronics and Communication
Basaveshwar Engineering College
Bagalkote, India.
Abstract:

The burgeoning integration of IoT and wearable technologies has revolutionized

healthcare, giving rise to a plethora of health monitoring systems. This survey
comprehensively analyzes the current landscape of these systems, examining their
architectures, sensor technologies, communication protocols, data management strategies,
and security measures. By identifying common challenges, such as data privacy, energy
efficiency, and real-time processing, this research underscores the need for innovative
solutions. The survey also explores the potential of emerging technologies like AI and
machine learning in enhancing health monitoring capabilities. Motivated by these findings,
we embarked on a work to develop a “Health Monitoring System using IoT” that addresses
specific identified gaps. This survey serves as a foundation for our project and guides future
research towards more efficient, accurate, and user-centric health monitoring solutions.

Keywords:

IoT, LCD, EEG, ECG, AI&M.

I Introduction:

The convergence of Internet of Things (IoT) and wearable technologies has ushered in a
new era of healthcare, characterized by the emergence of health monitoring systems. These
systems, equipped with an array of sensors including heart rate, temperature, humidity,
SpO2 (blood oxygen saturation), and potentially others such as blood pressure,
electrocardiogram (ECG), and electroencephalogram (EEG), can collect and transmitting
vital health data in real-time, enabling continuous and remote patient monitoring. By
providing insights into an individual's physiological state, these systems hold immense
potential to revolutionize preventive care, chronic disease management, and emergency
response.

384
 Survey on Health Monitoring System

The potential benefits of health monitoring systems are far-reaching. For patients, these
systems offer the convenience of self-management, empowering them to take proactive
steps towards improving their health. For healthcare providers, remote monitoring
capabilities enhance patient care, optimize resource allocation, and facilitate early
intervention. Additionally, health monitoring systems can contribute to population health
management by generating valuable data for epidemiological studies and public health
initiatives.

However, the development and deployment of effective health monitoring systems presents
significant challenges. Issues such as data privacy, security, energy efficiency, and
algorithm accuracy require careful consideration. Furthermore, the integration of these
systems into existing healthcare infrastructure necessitates a holistic approach that involves
collaboration between technologists, healthcare providers, and policymakers.

This survey aims to provide a comprehensive overview of the current state-of-the-art in

health monitoring systems, identifying key trends, challenges, and opportunities. By
examining existing research and development efforts, this study seeks to contribute to the
advancement of this rapidly evolving field and inform the design of future health monitoring
solutions.

II Literature Survey:

In this paper, the author Tarannum Khan & Manju K. Chattopadhyay [1] have proposed
smart health monitoring system effectively addresses the need for remote patient care,
especially in rural areas, by combining biomedical sensors, data storage, cloud connectivity,
and a mobile app. While successfully demonstrating functionality, the system can be
enhanced through improved sensor accuracy, data security, scalability, integration with
existing healthcare systems, advanced analytics, user-friendly design, expanded monitoring
parameters, and telemedicine capabilities. Long-term sustainability and cost-effectiveness
should also be considered for widespread implementation. Regular evaluation and updates
are essential to optimize the system's performance and impact on patient outcomes.

The author Srinivasarao Udaraet al. [2] discussed that their project aims to develop an
affordable, portable health monitoring system for remote areas with limited access to
specialist doctors. Utilizing IoT, the system enables data transmission to a remote server for
doctor access. It is built on an Arduino Mega 2560, featuring an ADS1292R ECG shield,
LM35 temperature sensor, ESP8266 Wi-Fi module, and a 16x2 LCD display. Sensors
provide analog outputs, interfaced with the Arduino's analog pins, while a Pulse oximeter
measures pulse rate and blood pressure. Data from these sensors is displayed locally on the
LCD and transmitted to an IoT platform via ESP8266. The platform logs and processes this
data, accessible from anywhere with the IoT account credentials. This system continuously
monitors patient conditions, sending alerts if critical thresholds are exceeded, and enables
remote specialists to review patient data. By consolidating various monitoring tools into one
module, the system simplifies patient monitoring in underserved areas.

In one of the works, the author Shubham Banka et.al.[3] proposed an IoT-based healthcare
system to enhance patient care, particularly in underserved areas. By utilizing wearable
devices and a Raspberry pi, the systems continuously monitor vital signs and transmit data
385
International Journal of Research and Analysis in Science and Engineering

to central server for analysis. This enables early disease detection. remote patient
monitoring & timely alerts to care givers. This system has potential to revolutionize health
care delivery by providing timely efficient & accessible care.

The author Preethi D. & Malathi M. [4] have discussed an IoT-based patient health
monitoring system utilizing an Arduino microcontroller to gather data from sensors like
pulse rate and temperature (potentially including blood pressure). This information is
displayed on an LCD, transmitted to a cloud platform for remote access and analysis, and
alerts are sent via buzzer and SMS/notification in case of abnormal readings. The work aims
to provide continuous, real-time patient monitoring, enabling timely medical interventions,
and empowering caregivers with accessible health updates. This technology holds the
potential to revolutionize healthcare by facilitating remote patient care, early detection of
health issues, and improved patient outcomes.

In this paper, the author Amol A. Sonune et. al. [5] discussed about distribution transformers
are crucial in power networks but require frequent monitoring due to their widespread
distribution. Traditional monitoring involves manual checks, which can be labour-intensive
and imprecise. This system leverages IoT to monitor transformer parameters such as load
current, voltage, oil level, and temperature. Using sensors, data is collected and sent to an
ESP32 microcontroller, which integrates Wi-Fi and Bluetooth, eliminating the need for
external modules and reducing the system's size. The ESP32 checks parameter limits and
sends data to the Blynk IoT platform, allowing operators to make informed decisions and
pre-empt failures. IoT enables remote sensing and control, improving efficiency and
reducing human intervention. While SCADA systems are used for power transformers, they
are costly and less practical for distribution transformers. This system provides real-time
monitoring, identifying potential issues like overloads and overheating, thus extending
transformer life and improving reliability. Embedded systems in this setup offer cost
savings, low power consumption, and high reliability, making it a practical alternative to
manual monitoring. The author Prajoona Valsalanet. al. [6] discussed about Health concerns
are crucial as technology advances, exemplified by the recent coronavirus impact on China's
economy. Remote health monitoring, powered by the Internet of Things (IoT), offers a
solution to manage epidemics effectively. This project focuses on developing a smart health
tracking system using sensors to monitor vital signs and internet connectivity to alert loved
ones of issues. The system aims to reduce healthcare costs by minimizing physician visits,
hospitalizations, and diagnostic tests. Sensors linked to a microcontroller display real-time
data on an LCD and send alerts via IoT if critical changes are detected. Compared to SMS-
based systems, IoT provides more detailed and accessible patient information through a web
interface. In rural areas, where medical facilities are scarce, this system enables early
detection of health issues, preventing severe complications and unnecessary costs. By
storing health data in the cloud and allowing remote access for doctors, IoT-based
monitoring enhances healthcare accessibility and efficiency.

In one of the work, the author Jayakumar S. et. al. [7] discussed that their project proposes
an IoT-based healthcare application for continuous, low-cost patient monitoring. The
system uses sensors like ECG monitors, heart rate monitors, and temperature sensors to
track vital signs. Data is collected and transmitted to the cloud via a microcontroller, where
it is processed and analysed. If an emergency is detected, alerts are sent to caregivers or
doctors, facilitating timely intervention. This approach addresses the challenge of limited
healthcare access in rural areas, where timely and affordable treatment is often lacking. The
386
 Survey on Health Monitoring System

system aims to provide vital parameters such as temperature and pulse using Thing Speak,
an IoT platform. The focus is on improving accuracy through IoT and enabling mass
screenings in remote areas. Future work will integrate machine learning for disease
prediction and enhance the system with lightweight sensors and improved features for
broader applicability.

The authors Sumeet Dhali & Suraj Prakash [8] proposed an IoT-based health monitoring
system which is developed to address the critical need for timely patient care. Wearable
sensors continuously monitor vital parameters, transmitting data to a central server for
analysis. This work enables remote patient monitoring, early disease detection, and
predictive healthcare. By providing real-time insights to healthcare professionals, the work
aims to improve patient outcomes and reduce mortality rates due to delayed treatment. The
integration of IoT technology has the potential to revolutionize healthcare delivery by
making it more accessible, efficient, and patient-centred. Additionally, this work can
facilitate personalized treatment plans, empower patients through self-management tools,
and contribute to the development of new healthcare models focused on prevention and
wellness.

The author Anil Kadu et. al. [9] have discussed the rise of user-friendly health monitoring
systems addressing a critical need timely and accurate diagnosis. Traditional methods
sometimes fail to detect diseases early, hindering proper treatment. This IoT-based system
tackles this challenge by continuously monitoring key patient parameters (pulse rate,
oxygen saturation) and predicting potential health issues. Sensor data is displayed on an
LCD and transmitted to an app, allowing for real-time tracking. Furthermore, the authors
also suggested to improve system's ability to generate predictive insights that can assist
healthcare providers in developing personalized treatment plans. This technology holds the
promise of transforming healthcare delivery into a more proactive and patient-centric
approach. Ultimately, this IoT-based health monitoring system represents a significant step
forward in preventing diseases, improving patient outcomes, and optimizing healthcare
resource utilization.

In this paper, the author Mohit Yadav et. al. [10] discussed about the advent of coronavirus,
healthcare has become a top priority globally. Patients often experience discomfort due to
bulky monitoring equipment with numerous wires. To address this, an IoT-based health
monitoring system is proposed as an effective solution. IoT, rapidly advancing in healthcare
research, enables remote monitoring of patients. This system allows authorized personnel
to track patients from remote locations and is particularly useful in overwhelmed hospitals
where emergency care is needed. The IoT-based system monitors vital signs such as heart
rate, body temperature, and more, predicting conditions like heart attacks or chronic
illnesses. It improves care quality by minimizing patient mobility and offering timely
assistance. Using Raspberry Pi, the system collects data from various sensors, including
temperature, pulse, and humidity, and transmits it to a cloud server. This data is then
accessible to doctors via a smartphone IoT platform, facilitating diagnosis and enhancing
patient care.

In one of the works, the author Mohammed Hasan Ali et. al. [11] discussed about monitoring
COVID-19 patients presents significant challenges, which can be addressed using IoT
technology. This research leverages IoT to track COVID-19 symptoms in real-time through
387
International Journal of Research and Analysis in Science and Engineering

wearable devices that monitor health indicators such as temperature, oxygen saturation, and
heart rate. The system comprises three layers: the first collects patient data, the second stores
it in the cloud, and the third uses the data to alert patients and their families. Deep-learning
models, specifically an optimized Convolutional Neural Network (MHCNN), are used for
further analysis and accurate classification of health conditions. IoT facilitates remote
monitoring, allowing for timely intervention and reducing the disease's spread by
identifying potential cases and managing contact tracing. Additionally, IoT can assist in
medication tracking, remote consultations, and even controlling drones for public health
measures. The system, tested with MATLAB, achieves 98.76% accuracy and aims for
continued improvements in disease recognition and accuracy.

The author Sharath M. et. al. [12] have addressed healthcare disparities in remote regions,
an IoT-based smart edge system is proposed. This work utilizes wearable sensors and
advanced algorithms to continuously monitor patient health, transforming raw data into
actionable health summaries and criticality alerts. By enabling early interventions, reducing
healthcare costs, and improving accessibility, this solution aims to enhance patient
outcomes, particularly in underserved areas. This work innovative approach has the
potential to revolutionize healthcare delivery by providing timely, data-driven care to those
who need it most. With the capacity to scale and integrate with existing healthcare systems,
this IoT solution can contribute to building a more equitable and efficient healthcare
ecosystem. In this paper, the author M. Malathi et. al. [13] discussed about Health
monitoring systems are evolving rapidly, addressing the challenge of continuous patient
monitoring in hospitals. Their project uses an Arduino Mega 2560 microcontroller to track
vital signs such as blood pressure, body temperature, and heart rate. Embedded C is
employed to process sensor data, which is then transmitted via GSM and Wi-Fi modules to
update both caretakers and doctors through IoT cloud storage and web pages. The system
is particularly relevant during the COVID-19 pandemic, reducing the need for in-person
consultations and maintaining social distancing. It also includes an ECG sensor to monitor
heart activity. The IoT-based system offers significant cost savings, enhances patient care,
and can be applied in emergency situations such as road accidents. It reduces the need for
constant human oversight, making it ideal for bedridden patients and those in ICU. The
system operates with 90% accuracy and plans to integrate additional sensors like cholesterol
and glucose monitors in the future.

III Conclusion:

In conclusion, health monitoring systems, powered by IoT and wearable technologies, have
immense potential to revolutionize healthcare delivery. This survey has unveiled the current
state-of-the-art in these systems, highlighting their architectural diversity, sensor
capabilities, communication protocols, and data management strategies. While significant
strides have been made, challenges such as data privacy, energy efficiency, and real-time
processing persist. The integration of artificial intelligence and machine learning offers
promising avenues for addressing these challenges and unlocking new possibilities for early
disease detection, personalized treatment plans, and remote patient monitoring. By
addressing these critical areas, we can harness the full potential of health monitoring
systems to improve population health outcomes and enhance the quality of life for
individuals worldwide.

388
 Survey on Health Monitoring System

References:

1. Tarannum Khan and Manju K. Chattopadhyay, “SMART HEALTH MONITORING

SYSTEM”, IEEE, International Conference on Information, Communication,
Instrumentation and Control (ICICIC - 2017), Paper id :387
2. Yedukondalu Udara, Srinivasarao Udara, Harish H M, Hadimani H C, “HEALTH
MONITORING SYSTEM USING IOT”, International Journal of Engineering and
Manufacturing Science. ISSN 2249-3115 Volume 8, Number 1 (2018) pp. 177-182
3. Shubham Banka, Isha Madan and S.S. Saranya, “Smart Healthcare Monitoring using
IoT”, International Journal of Applied Engineering Research ISSN 0973-4562 Volume
13, Number 15 (2018) pp. 11984-11989
4. Malathi M, Preethi D, “IoT based Patient Health Monitoring System”, International
Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 RTICCT -
2019
5. Amol A. Sonune, Mayur S. Talole, Sumit V. Sonkusale, Vivek N. Gayki, Piyush S.
Jaiswal, Ankit A. Akotkar, “Condition Monitoring of Distribution Transformer using
IOT”, International Journal of Engineering Research & Technology (IJERT) ISSN:
2278-0181 IJERTV9IS060244 Vol. 9 Issue 06, June-2020
6. Prajoona Valsalan, Tariq Ahmed Barham Baomar, Ali Hussain Omar Baabood, “IOT
BASED HEALTH MONITORING SYSTEM”, Journal of Critical reviews, ISSN-
2394-5125, Vol 7, Issue 4, 2020
7. Jayakumar S, Ranjithkumar R, Tejswini R and Kavil S, “IoT Based Health Monitoring
System”,the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC
4.0), doi:10.3233/APC210140
8. Sumeet Dhali and Suraj Prakash, “SMART PATIENT HEALTH MONITORING
SYSTEM USING IOT”, 2021
9. Anil Kadu, Mansi Gajare, Pritesh Chaudhari, Payal Deshmukh, Sameer Deshmukh,
Manas Dani, “IOT BASED HEALTH MONITORING SYSTEM”, Conference: IOT
Based Health Monitoring System, January 2021
10. Mohit Yadav, Aditya Vardhan, Amarjeet Singh Chauhan, Sanjay Saini, “IOT BASED
HEALTH MONITORING SYSTEM”, 2022 IJCRT | Volume 10, Issue 1 January 2022
| ISSN: 2320-2882
11. Mustafa Musa Jaber, Thamer Alameri, Mohammed Hasan Ali, Adi Alsyouf,
Mohammad Al-Bsheish, Badr K, Aldhmadi, Sarah Yahya Ali, Sura Khalil Abd, Saif
Mohammed Ali, Waleed Albaker and Mu’taman Jarrar, “REMOTELY MONITORING
COVID-19 PATIENT HEALTH CONDITION USING METAHEURISTICS
CONVOLUTE NETWORKS FROM IOT-BASED WEARABLE DEVICE HEALTH
DATA”, DOI: 10.3390/s22031205, Feb 2022
12. Sharath M, Saran V, Pradeep K, Rudrasamy K, “PATIENT HEALTH MONITORING
SYSTEM USING IOT”, 2022 IJCRT | Volume 10, Issue 6 June 2022 | ISSN: 2320-
2882
13. Malathi. M, Muniappan, Praveen Kumar Misra, Balaji Ramkumar Rajagopal, Pankaj
Borah, “A SMART HEALTHCARE MONITORING SYSTEM FOR PATIENTS
USING IOT AND CLOUD COMPUTING”, International Conference on Biomedical
Engineering and Computing Technologies (ICBECT ’22) AIP Conf. Proc. 2603,
030012-1–030012-6.

389
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

41. Towards 6G: An Overview of Next Generation

Communication
Mallikarjun Dheshmuk, Achyut Yaragal
Research Scholar, Dept. of Electronics and Communication,
Basaveshwar Engineering College, Bagalkote.
S. B. Kumbalavati, Kirankumar Y. Bendigeri,
Jayashree D. Mallapur
Dept. of Electronics and Communication,
Basaveshwar Engineering College, Bagalkote
Abstract:

Sixth generation (or 6G) cellular communication standards are a major improvement over
the existing 5G networks. As we approach the dawn of this new era, 6G promises to bring
about significant changes that will enhance connection and open up a plethora of creative
applications in a variety of fields. The transition from 1G to 5G has already altered how we
work, live, and communicate with one another. This platform will be enhanced by 6G, which
will offer hitherto unheard-of capabilities. Through this paper, various research areas of
6G are discussed.

Keywords:

6G, IRS, AI & Machine learning, OTFS.

I Introduction:

6G, or the sixth generation of wireless communication technology, is the anticipated

successor to 5G. While still in the research and development phase, 6G is expected to
revolutionize mobile communications by offering unprecedented speed, connectivity, and
capabilities. The following are some of the Key Features of 6G:

A. Higher Data Rates:

Data rates of up to one terabit per second (Tbps) are anticipated with 6G, significantly
outpacing 5G's gigabit speeds [1,2]. This will facilitate the real-time distribution of massive
volumes of data, supporting applications that require a lot of data, such as augmented reality
and HD video streaming.

390
 Towards 6G: An Overview of Next Generation Communication

B. Ultra-Low Latency:

6G is notable for its extremely low latency, which can be as low as a few milliseconds [3].
Applications requiring quick response times, such as industrial automation, remote surgery,
and autonomous driving, will require this.

C. Enhanced Connectivity:

From densely populated urban areas to sparsely populated remote rural locations, 6G
promises to deliver flawless connectivity in a variety of settings. A ubiquitous system is one
of the proposals of 6G wherein terrestrial, satellite, underwater are brought together on a
common platform [4].

D. Advanced AI and Machine Learning Integration:

Self-optimized network and management can be achieved by the incorporation of AI ML

techniques in 6G [5]. AI-ML technologies will enhance security standards, manage
resources more effectively, and maximize network performance.

E. Improved Security and Privacy:

With the increasing reliance on digital communication, security and privacy are of outmost
importance. 6G will leverage advanced encryption techniques and AI-driven security
techniques to protect data and ensure user secrecy [6].

Figure 1: Features of 6G

391
International Journal of Research and Analysis in Science and Engineering

II Technological Innovations:

A. Terahertz (THz) Frequency Bands):

Utilizing the THz spectrum (0.1-10 THz) is of outmost importance for achieving the high
data rates of 6G. Research is ongoing to design THz antennas, propagation models and
transceivers. Innovations in materials science and photonics will be critical in overcoming
challenges like atmospheric attenuation and signal dispersion[7].

B. Advanced Modulation Techniques:

Recent modulation methods such as Quadrature Amplitude Modulation (QAM) and Filtered
Orthogonal Frequency Division Multiplexing (F-OFDM) are being explored to increase
spectral utilization. These techniques will enable faster data rates over the same bandwidth,
crucial for meeting the data demands of future applications.

C. Intelligent Reflecting Surfaces (IRS):

IRS essentially involves using large number of small surfaces equipped with adjustable
elements to reflect signals in desired directions, enhancing coverage and capacity [8-9]. This
can significantly decrease the need for extra base stations, making network deployment
more cost-effective and energy-efficient.

D. Quantum Communications:

Quantum technologies, particularly quantum key distribution (QKD), aims to revolutionize

security of network [10]. Quantum superposition and entanglement principles enable
unbreakable encryption, ensuring data confidentiality and secrecy against future cyber
threats.

E. Network Slicing and Edge Computing:

Network slicing creates virtual networks which can be customized to specific services,
ensuring optimal performance for different applications. Edge computing will process data
nearer to the source, thus reducing the latency and bandwidth usage, and enabling real-time
data analytics and decision-making.

F. NOMA:

Non-Orthogonal Multiple Access." It's a technology designed to improve the efficiency and
capacity of wireless communication systems. NOMA is a multiple access technique that
allows multiple users to share the same resource blocks simultaneously [11]. Unlike
traditional Orthogonal Multiple Access (OMA) methods, NOMA differentiates users by
their power levels or code, not by time or frequency. Enhance spectral efficiency,
accommodate a large number of devices, and improve user experience by supporting diverse
service requirements. Power Domain NOMA: Users are allocated varying power levels,
depending on the distance.

392
 Towards 6G: An Overview of Next Generation Communication

The base station transmits a superimposed signal with different power levels, and users
reconstruct their respective signals using a technique known as Successive Interference
Cancellation (SIC). Code Domain NOMA: Users are assigned different codes. Signals are
superimposed in the code domain, and users separate their signals using specific decoding
techniques.

Figure 2: NOMA

G. OTFS:

OTFS (Orthogonal Time Frequency Space) modulation is a next generation modulation

technique aimed to overcome the issues of high mobility and multipath propagation in
wireless communication systems. The transmission symbols are converted in to a two-
dimensional delay Doppler spread domain [12]. OTFS has the capability to provide high
speed mobility support because of delay Doppler domain. Conventionally as the speed of
UE increases, it becomes difficult to estimate the rapidly changing channel characteristics
and hence high-speed mobility is an issue in prior generation communication. This method
provides robustness against interference and changes in the channel. This feature of 6G is
particularly very useful for applications like high-speed bullet trains, drones, UAV, aero
planes etc.

• Delay-Doppler Domain: OTFS operates in delay Doppler domain compared to earlier

techniques which operated in time frequency domain. It captures and models the effect
of Doppler more precisely and helps in robust communication.
• Channel Resilience: By transforming complex channel impairments like fading and
Doppler effects into more controllable forms in the Delay-Doppler domain, OTFS is
able to handle severe channel impairments.
• Improved Detection: In dynamic situations, the modulation approach makes it easier
to detect and decode information.
• High Mobility Support: Excels in situations when the transmitter and receiver are
moving at high relative speeds, like high-speed trains and vehicular communication.
• Enhanced Spectral Efficiency: Better utilization of the spectrum because of its strong
multipath and Doppler impact handling.
393
International Journal of Research and Analysis in Science and Engineering

• Better Performance in Multipath Channels: handles multipath propagation's

difficulties well, increasing communication dependability overall.

Figure 3: Applications

III Applications of 6G:

The following are some of the applications of 6G [13].

• Autonomous Driving: Enhancing the communication capabilities of autonomous

vehicles, enabling real-time data exchange with other vehicles, infrastructure, and
pedestrians.
• Traffic Management: Improving traffic flow, reducing congestion, and enhancing
road safety through intelligent traffic management systems.
• Extended Reality (XR) VR and AR: Providing the bandwidth and low latency
required for seamless VR and AR experiences, transforming education, training,
healthcare, and entertainment.
• Remote Collaboration: Enabling realistic and interactive remote collaboration
environments, enhancing productivity and engagement in virtual meetings and
conferences.
• Smart Cities and Infrastructure: IoT Integration: Supporting extensive IoT networks
for smart utilities, waste management, and energy systems, enhancing urban living and
sustainability.
• Advanced Surveillance: Facilitating real-time, high-definition surveillance and
monitoring systems for public safety and smart infrastructure management.
• Healthcare and Telemedicine: Remote Surgery: Enabling high-definition, real-time
remote surgeries with minimal latency, improving access to medical expertise in remote
areas [14].

394
 Towards 6G: An Overview of Next Generation Communication

• Telehealth: Enhancing telehealth services with high-resolution video consultations,

remote monitoring, and AI-driven diagnostics.
• Industrial Automation and Robotics: Smart Manufacturing: Supporting advanced
manufacturing processes with real-time control and automation, improving efficiency
and reducing downtime [15].
• Collaborative Robots: Enabling robots to work safely and efficiently alongside
humans in various industrial and service applications.
• Entertainment and Media: Immersive Experiences: Delivering high-quality,
immersive content with augmented reality, holography, and interactive media
experiences. Interactive Gaming: Enhancing online gaming with low latency, high
resolution, and seamless multiplayer experiences.
• Climate Monitoring: Supporting networks of sensors for real-time climate monitoring
and disaster management, enhancing environmental protection efforts.
• Sustainable Development: Enabling smart grids, efficient energy management, and
sustainable resource utilization across various industries.
• Satellite Communication: Enhancing satellite communication systems, providing
high-speed, low-latency connectivity in remote and underserved regions.
• Space Exploration: Supporting advanced communication systems for space missions,
enabling real-time data transmission and control for space exploration.

IV Challenges and Considerations:

A. Technical Challenges:

Developing terahertz technology, making sure artificial intelligence is seamlessly integrated

into network management, and reducing the complexity of network slicing are important
problems. The challenges of signal transmission, spectrum assignment, and interference
mitigation in the THz range require research.

B. Regulatory and Standardization Issues:

Worldwide standards must be established for 6G in order to ensure interoperability and a

smooth worldwide deployment. In order to distribute spectrum, establish technological
procedures, and handle cybersecurity laws, authorities will need to work together. In order
to prevent fragmentation and provide a standard 6G environment, international cooperation
will be required.

C. Security and Privacy Concerns:

As the communication system grows more intricate every day, maintaining privacy and
secrecy will become increasingly difficult. To protect user data and uphold confidence,
advanced encryption standards, AI-driven security algorithms, and extensive privacy
regulations will be required.

395
International Journal of Research and Analysis in Science and Engineering

D. Environmental Impact:

Sustainable solutions are needed for sixth generation infrastructure's power consumption
and e-waste production. Reusable tactics, green system design, and clever energy-saving
measures will be essential components in lowering 6G networks' environmental impact.

V Future Prospects OF 6G:

A. Expected Timeline for Deployment:

It is projected that the first commercial 6G systems will go live in 2030 after extensive
standardization, development, and research efforts. To prepare for this upgrade, test
frameworks and prototype projects are already being established globally.

B. Potential Economic and Societal Impacts:

It is projected that 6G will significantly accelerate economic growth by opening up new

markets, fostering innovation, and raising productivity in a variety of industries. It will also
have a significant impact on society, opening up new avenues for education, entertainment,
and healthcare as well as enhancing general quality of life and closing the digital divide.

C. Research and Development Directions:

Subsequent investigations will concentrate on reducing the technological obstacles,

developing novel technologies, and investigating intelligent uses. Public private partnership
mode, international agreement, and collaborative research projects will be necessary to
enhance 6G technology and guarantee its successful global rollout.

VI Conclusion:

As the 6G age approaches, there are a ton of fascinating and huge possibilities. In addition
to being a 5G upgrade, 6G will bring about a radical change in computation, networking,
and communication. We can fully utilize 6G by resolving its technical, legal, and
environmental issues, opening the door to a more intelligent, connected, and sustainable
future. The development of 6G is a joint project, and its effective execution will require
cooperation, ingenuity, and a dedication to conquering the obstacles that lie ahead. Through
this paper, various aspects of 6G are discussed, which helped in narrowing down objectives.
Going ahead, the future research work will focus on intelligent reconfigurable surfaces,
NOMA, OTFS.

References:

1. W. Jiang, B. Han, M. A. Habibi, and H. D. Schotten, “The road towards 6G: A

comprehensive survey,” IEEE Open J. Commun. Soc., vol. 2, pp. 334–366, 2021.
2. B. Zong, X. Duan, C. Fan and K. Guan, "6G Technologies - Opportunities and
Challenges," 2020 IEEE International Conference on Integrated Circuits, Technologies
and Applications (ICTA), Nanjing, China, 2020, pp. 171-173.

396
 Towards 6G: An Overview of Next Generation Communication

3. S. Dang, O. Amin, B. Shihada, and M.-S. Alouini, “What should 6G be?” Nat.
Electron., vol. 3, no. 1, pp. 20–29, 2020.
4. W. Saad, M. Bennis, and M. Chen, “A vision of 6G wireless systems: Applications,
trends, technologies, and open research problems,” IEEE Netw., vol. 34, no. 3, pp. 134–
142, 2019.
5. I. F. Akyildiz, A. Kak, and S. Nie, “6G and beyond: The future of wireless
communications systems,” IEEE Access, vol. 8, pp. 133995–134030, 2020.
6. P. Porambage, G. Gur, D. P. Moya Osorio, M. Livanage, and M. Ylianttila, “6G
Security Challenges and Potential Solutions,” in 2021 Joint European Conference on
Networks and Communications & 6G Summit (EuCNC/6G Summit), Porto, Portugal:
IEEE, Jun. 2021, pp. 622–627.
7. T. S. Rappaport et al., “Wireless communications and applications above 100 GHz:
Opportunities and challenges for 6G and beyond,” IEEE Access, vol. 7, pp. 78729–
78757, 2019.
8. S. P. Dash and A. Kaushik, “RIS-Assisted 6G Wireless Communications: A Novel
Statistical Framework in the Presence of Direct Channel,” IEEE Commun. Lett., vol.
28, no. 3, pp. 717–721, Mar. 2024.
9. Z. Zhang et al., "Active RIS vs. Passive RIS: Which Will Prevail in 6G?" in IEEE
Transactions on Communications, vol. 71, no. 3, pp. 1707-1725, March 2023
10. S. J. Nawaz, S. K. Sharma, S. Wyne, M. N. Patwary and M. Asaduzzaman, "Quantum
Machine Learning for 6G Communication Networks: State-of-the-Art and Vision for
the Future," in IEEE Access, vol. 7, pp. 46317-46350, 2019.
11. Srivastava and P. P. Dash, “Non-Orthogonal Multiple Access: Procession towards B5G
and 6G,” in 2021 IEEE 2nd International Conference on Applied Electromagnetics,
Signal Processing, & Communication (AESPC), Nov. 2021, pp. 1–4.
12. S. S. Das and R. Prasad, “OTFS: Orthogonal Time Frequency Space Modulation A
Waveform for 6G,” in OTFS: Orthogonal Time Frequency Space Modulation A
Waveform for 6G, River Publishers, 2021, pp. i–xxvi. Accessed: Jul. 31, 2024.
13. M. Z. Chowdhury, M. Shahjalal, S. Ahmed and Y. M. Jang, "6G Wireless
Communication Systems: Applications, Requirements, Technologies, Challenges, and
Research Directions," in IEEE Open Journal of the Communications Society, vol. 1, pp.
957-975, 2020.
14. J. Omkar, S. Menaka, J. Praveena, N. Varshney, B. Arunkumar and G. V. Reddy, "The
Use of 6G Communication Technology in Healthcare Applications for the Accurate and
Better Transmission," 2024 4th International Conference on Advance Computing and
Innovative Technologies in Engineering (ICACITE), Greater Noida, India, 2024, pp.
1993-1999.
15. M. Yadav et al., "Exploring Synergy of Blockchain and 6G Network for Industrial
Automation," in IEEE Access, vol. 11, pp. 137163-137187, 2023, doi:
10.1109/ACCESS.2023.3338861.

397
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

42. A Comprehensive Review of Resource

Management Techniques in Edge Computing
Shrishial N. Chamatagoudar
Walchand Institute of Technology,
Solapur, Maharashtra, India.
Rajani S. Pujar, Mahabaleshwar S. K.,
Chayalakshmi C. L.
Basaveshwar Engineering College,
Bagalkote, Karnataka, India.

ABSTRACT:

Edge computing is a transformative approach that brings computation and data storage
closer to the data source, enhancing response times and saving bandwidth. This paper
provides a comprehensive review of resource management techniques in edge computing,
addressing the key challenges and presenting future directions for research. We discuss
various methods for resource allocation, load balancing, scheduling, and optimization. The
review also highlights the security aspects and potential solutions for efficient resource
management in edge environments.

I Introduction:

Edge computing is a transformative paradigm that represents a significant shift from

traditional cloud computing architectures. Unlike cloud computing, which relies on
centralized data centers, edge computing brings computation and data storage closer to the
data sources, i.e., to the 'edge' of the network. This proximity to data sources offers
numerous benefits, including reduced latency, improved bandwidth efficiency, enhanced
privacy, and greater reliability. The impetus for edge computing stems from the explosive
growth of data generated by Internet of Things (IoT) devices. According to a report by
Cisco, the number of connected devices is expected to reach 29.3 billion by 2023, generating
an unprecedented amount of data. Traditional cloud computing models are not well-suited
to handle this deluge of data due to inherent latency and bandwidth limitations. By
processing data closer to where it is generated, edge computing addresses these challenges
effectively. Resource management is a critical aspect of edge computing. Effective resource
management ensures that computational resources are allocated efficiently, tasks are
scheduled optimally, and loads are balanced across the network. This is particularly
important in edge computing environments due to the heterogeneous and distributed nature

398
 A Comprehensive Review of Resource Management Techniques in Edge Computing

of edge devices. Unlike homogeneous data center environments, edge computing

environments consist of a diverse array of devices with varying computational capacities,
connectivity options, and power constraints. This diversity necessitates advanced resource
management techniques that can adapt to the unique characteristics of edge environments.

Resource allocation in edge computing involves assigning computational resources to tasks

in a manner that maximizes performance while minimizing latency and energy
consumption. Traditional resource allocation algorithms, such as those used in cloud
computing, are not directly applicable to edge computing due to the dynamic and distributed
nature of edge environments. As a result, novel algorithms that consider factors such as
device heterogeneity, dynamic workloads, and real-time constraints are required. For
instance, machine learning-based algorithms have been proposed to predict resource
requirements and optimize allocation dynamically (Shi et al., 2016). Load balancing in edge
computing aims to distribute workloads evenly across available resources to prevent any
single node from becoming a bottleneck. Effective load balancing improves system
performance, enhances reliability, and ensures a more efficient utilization of resources.
Various strategies have been proposed for load balancing in edge computing, including
round-robin, least connections, and dynamic load balancing methods. These strategies differ
in their complexity, adaptability, and effectiveness. For example, dynamic load balancing
methods use real-time data to make informed decisions about workload distribution, thereby
adapting to changing network conditions and workload characteristics (Zhang et al., 2018).
Resource scheduling in edge computing determines the order and timing of task execution.
Scheduling tasks efficiently is crucial for meeting the quality of service (QoS) requirements
of edge applications, particularly those with real-time constraints. Real-time scheduling
algorithms prioritize tasks based on their urgency and deadlines, ensuring that critical tasks
are executed promptly. Non-real-time scheduling algorithms, on the other hand, focus on
optimizing overall system throughput and resource utilization. Hybrid scheduling
approaches that combine elements of both real-time and non-real-time scheduling have also
been explored to balance the competing demands of different types of applications (Xu et
al., 2020).

Optimization techniques in edge computing involve adjusting resource usage to improve

performance and efficiency. These techniques include load prediction models, which
anticipate future workloads based on historical data and adjust resource allocation
proactively. Additionally, energy-efficient algorithms aim to minimize energy consumption
without compromising performance, which is particularly important for battery-powered
edge devices. For instance, energy-aware task offloading strategies have been developed to
reduce energy consumption by dynamically deciding whether to process tasks locally or
offload them to more powerful edge servers (Wang et al., 2019).

The objectives and scope of this review are to provide an in-depth analysis of resource
management techniques in edge computing, with a focus on recent advancements and
challenges. The paper is structured as follows: Section 2 presents the background and
related work, providing a comprehensive overview of existing surveys and studies in the
field. Section 3 discusses various resource management techniques, including resource
allocation, load balancing, scheduling, and optimization, highlighting key algorithms and
methodologies. Section 4 addresses the challenges and future directions in resource
management for edge computing, emphasizing areas that require further research.
399
International Journal of Research and Analysis in Science and Engineering

Finally, Section 5 concludes the paper by summarizing the key findings and discussing their
implications for future research and practice.

II Literature Survey:

Edge computing has emerged as a crucial paradigm in the realm of distributed computing,
aiming to mitigate the limitations of traditional cloud computing by bringing computational
resources closer to the data sources. This section delves into the foundational concepts of
edge computing, explores its evolution, and examines related work in the field. Edge
computing refers to a distributed computing framework where data processing occurs at the
edge of the network, near the source of data generation. Unlike centralized cloud computing,
which relies on remote data centers, edge computing leverages local processing units, such
as gateways, routers, and end devices. This proximity reduces latency, conserves
bandwidth, and enhances the responsiveness of applications. The concept of edge
computing is not entirely new. It evolved from earlier paradigms like content delivery
networks (CDNs) and fog computing. CDNs aimed to reduce latency by caching content
closer to users, while fog computing, introduced by Cisco in 2012, extended cloud
capabilities to the network edge, providing a foundation for today's edge computing models.

Numerous surveys have been conducted to explore different facets of edge computing. For
instance, Shi et al. (2016) provided a comprehensive overview of edge computing,
highlighting its potential applications and challenges. They discussed how edge computing
complements cloud computing by addressing issues related to latency, bandwidth, and
privacy. Similarly, Satyanarayanan (2017) emphasized the importance of edge computing
in enabling real-time applications, particularly in the context of IoT. In the domain of
resource management, several surveys have focused on specific aspects such as resource
allocation, load balancing, and scheduling. For example, Abbas et al. (2018) reviewed
resource management techniques in mobile edge computing, categorizing them into
resource provisioning, offloading, and orchestration. Their survey underscored the need for
adaptive and scalable resource management solutions to handle the dynamic nature of edge
environments.

While existing surveys provide valuable insights, this review aims to fill certain gaps by
offering a more focused and detailed analysis of resource management techniques in edge
computing. Specifically, it addresses the following aspects:

• Comprehensive Coverage: Unlike previous surveys that focus on individual aspects

of resource management, this review encompasses a broader range of techniques,
including resource allocation, load balancing, scheduling, and optimization. By
providing a holistic view, it aims to highlight the interplay between these techniques
and their collective impact on system performance.
• Recent Advancements: Edge computing is a rapidly evolving field, with new
techniques and approaches being proposed regularly. This review incorporates recent
advancements from the past few years, ensuring that readers are updated with the latest
trends and innovations.
• Challenges and Future Directions: While previous surveys often conclude with a
discussion of challenges, this review delves deeper into specific issues such as security,
scalability, and interoperability. It also outlines potential future research directions,
400
 A Comprehensive Review of Resource Management Techniques in Edge Computing

offering a roadmap for researchers and practitioners to address current limitations and
explore new opportunities.

By addressing these key differences and contributions, this review aims to provide a
valuable resource for researchers, developers, and decision-makers involved in the design
and deployment of edge computing systems.

III Resource Management in Edge Computing:

• Resource management in edge computing is a multifaceted challenge that encompasses

several interrelated aspects, including resource allocation, load balancing, scheduling,
and optimization. This section explores each of these aspects in detail, highlighting their
importance, techniques, and challenges.

o Resource Allocation; Resource allocation involves assigning computational resources

to tasks in a manner that maximizes performance and efficiency. In edge computing,
this task is complicated by the heterogeneity and dynamic nature of edge devices.
Traditional resource allocation algorithms used in cloud computing are often unsuitable
for edge environments, necessitating the development of novel approaches.
o Techniques and Algorithms:
o Heuristic Algorithms: Heuristic algorithms, such as greedy algorithms and genetic
algorithms, provide near-optimal solutions for resource allocation problems. These
algorithms are particularly useful in edge computing due to their ability to handle
complex and dynamic environments. For example, a greedy algorithm might allocate
resources based on the current availability and task requirements, while a genetic
algorithm could evolve resource allocation strategies over time to optimize
performance.
o Machine Learning-based Algorithms: Machine learning-based algorithms leverage
historical data and predictive models to optimize resource allocation dynamically.
These algorithms can adapt to changing conditions and workload patterns, making them
well-suited for edge environments. Techniques such as reinforcement learning and deep
learning have been applied to resource allocation problems, with promising results. For
instance, a reinforcement learning algorithm might learn to allocate resources based on
feedback from previous allocations, continually improving its performance over time.

Resource allocation in edge computing faces several challenges, including resource

heterogeneity, dynamic network conditions, and varying workload demands. Addressing
these challenges requires adaptive and scalable solutions that can respond to changes in
real-time. For instance, adaptive algorithms can adjust resource allocation based on real-
time monitoring data, ensuring optimal performance even in dynamic environments.

• Load Balancing: Load balancing distributes workloads evenly across available

resources to prevent any single node from becoming a bottleneck. Effective load
balancing improves system performance, enhances reliability, and ensures efficient
utilization of resources.

o Round-Robin: Round-robin is a simple and effective load balancing strategy for

homogeneous environments. It distributes tasks in a cyclic manner, ensuring that all
401
International Journal of Research and Analysis in Science and Engineering

nodes receive an equal share of the workload. This strategy is easy to implement and
works well when all nodes have similar capabilities.
o Least Connections: Least connections is suitable for environments with varying
workload sizes. It assigns tasks to the node with the fewest active connections,
balancing the load based on current activity levels. This strategy is more adaptable than
round-robin and can handle heterogeneous environments more effectively.
o Dynamic Load Balancing: Dynamic load balancing methods use real-time data to
make informed decisions about workload distribution. These methods can adapt to
changing conditions and workload patterns, ensuring optimal performance. For
example, a dynamic load balancing algorithm might monitor resource usage across the
network and redistribute tasks as needed to prevent overloads and ensure efficient
utilization of resources.

• Resource Scheduling: Resource scheduling determines the order and timing of task
execution. Efficient scheduling is crucial for meeting the quality of service (QoS)
requirements of edge applications, particularly those with real-time constraints.

o Priority-based Scheduling: Priority-based scheduling assigns tasks based on their

priority levels. High-priority tasks are executed first, ensuring that critical applications
receive the necessary resources. This approach is particularly useful for applications
with stringent QoS requirements, such as real-time video streaming or autonomous
vehicle control.
o Fair Scheduling: Fair scheduling ensures equal resource allocation to all tasks,
promoting fairness and preventing resource monopolization. This approach is suitable
for environments where all tasks have similar importance and resource demands. For
example, a fair scheduling algorithm might allocate resources evenly among all tasks,
ensuring that no single task dominates the available resources.
o Real-time vs. Non-real-time Scheduling: Real-time scheduling is crucial for
applications with strict latency requirements, such as industrial automation or
telemedicine. Real-time scheduling algorithms prioritize tasks based on their urgency
and deadlines, ensuring that critical tasks are executed promptly. Non-real-time
scheduling algorithms, on the other hand, focus on optimizing overall system
throughput and resource utilization, making them more suitable for applications with
less stringent timing requirements.

• Resource Monitoring and Optimization: Resource monitoring involves tracking

resource usage to ensure efficient utilization. Optimization techniques adjust resource
usage to improve performance and efficiency.

o Agent-based Monitoring: Agent-based monitoring deploys agents to collect resource

usage data from various nodes in the network. These agents can provide detailed and
real-time insights into resource utilization, helping to identify bottlenecks and optimize
resource allocation. For example, an agent might monitor CPU and memory usage on
an edge device, reporting this data to a central controller for analysis and optimization.
o Network-wide Monitoring: Network-wide monitoring uses network devices to
monitor resource usage across the entire edge network. This approach provides a
comprehensive view of resource utilization and helps to identify patterns and trends.

402
 A Comprehensive Review of Resource Management Techniques in Edge Computing

For instance, a network-wide monitoring system might track data traffic and processing
loads across multiple edge nodes, providing insights into overall network performance.
o Optimization Techniques:
o Load Prediction Models: Load prediction models anticipate future workloads based
on historical data and adjust resource allocation proactively. These models can help
prevent resource shortages and ensure optimal performance. For example, a load
prediction model might analyze past usage patterns to predict future demand, allowing
resources to be allocated in advance to meet anticipated needs.
o Energy-efficient Algorithms: Energy-efficient algorithms aim to minimize energy
consumption while maintaining performance, which is particularly important for
battery-powered edge devices. These algorithms optimize resource usage to reduce
energy consumption, extending the battery life of edge devices. For instance, an energy-
efficient algorithm might offload tasks to more energy-efficient nodes or adjust
processing speeds to balance performance and energy use.

By addressing these key aspects of resource management, this section provides a

comprehensive overview of the techniques and challenges involved in optimizing resource
usage in edge computing environments.

IV Challenges and Future Directions:

The field of edge computing presents several challenges that need to be addressed to realize
its full potential. This section explores the key challenges and outlines potential future
research directions.

• Security Issues: Security is a significant concern in edge computing due to the

distributed nature of edge devices. These devices are often located in less secure
environments, making them vulnerable to physical attacks and unauthorized access.
Additionally, the data processed at the edge may be sensitive, necessitating robust
security measures to protect against data breaches and cyberattacks.
• Resource Heterogeneity: Edge computing environments consist of a diverse array of
devices with varying computational capacities, storage capabilities, and energy
resources. Managing this heterogeneity poses a challenge for resource allocation, load
balancing, and scheduling. Ensuring that tasks are executed efficiently across different
devices requires adaptive and scalable resource management solutions.
• Dynamic Network Conditions: The network conditions in edge computing
environments are highly dynamic, with varying levels of connectivity, bandwidth, and
latency. These fluctuations can impact the performance and reliability of edge
applications, necessitating real-time monitoring and adaptive resource management
strategies.

Emerging Trends and Future Research Directions:

• AI and Machine Learning Integration: Integrating artificial intelligence (AI) and

machine learning (ML) techniques into edge computing can enhance predictive
resource management, improve decision-making, and optimize performance. For
example, ML algorithms can analyze historical data to predict future workloads and
dynamically adjust resource allocation to meet demand.
403
International Journal of Research and Analysis in Science and Engineering

• Security Enhancements: Developing robust security frameworks for edge computing

is critical to protect against cyber threats and ensure data privacy. Future research
should focus on advanced encryption techniques, secure communication protocols, and
intrusion detection systems tailored for edge environments.
• Scalability Solutions: As the number of edge devices continues to grow, ensuring that
resource management techniques can scale effectively is crucial. Future research should
explore scalable algorithms and architectures that can handle large-scale deployments
while maintaining performance and reliability.
• Interoperability: Ensuring interoperability between different edge computing
platforms and devices is essential for seamless integration and operation.
Standardization efforts and the development of common protocols can facilitate
interoperability and promote the widespread adoption of edge computing.
• Energy Efficiency: Optimizing energy consumption is a critical concern, particularly
for battery-powered edge devices. Future research should focus on developing energy-
efficient algorithms, exploring alternative energy sources, and designing hardware that
reduces power consumption. By addressing these challenges and exploring these future
research directions, the field of edge computing can continue to evolve and expand,
unlocking new opportunities and applications.

V Conclusion:

This paper provided a comprehensive review of resource management techniques in edge

computing. We discussed various methods for resource allocation, load balancing,
scheduling, and optimization. Future research should focus on integrating AI, enhancing
security, and developing scalable solutions for efficient resource management in edge
environments.

References:

1. S. K. Garg et al., 'Resource Management Approaches in Fog Computing: A

Comprehensive Review,' Future Generation Computer Systems, vol. 93, pp. 412-427,
2019. DOI: 10.1007/S10723-019-09491-1.
2. M. Abdelshkour et al., 'Resource Management in Fog/Edge Computing: A Survey on
Architectures, Infrastructure, and Algorithms,' IEEE Access, vol. 8, pp. 21429-21442,
2020. DOI: (missing DOI).
3. P. McGrath et al., 'Resource Management Techniques for Cloud/Fog and Edge
Computing: An Evaluation Framework and Classification,' Sensors, vol. 21, no. 5, pp.
1832, 2021. DOI: 10.3390/S21051832.
4. L. Zhang et al., 'A Full Dive into Realizing the Edge-Enabled Metaverse: Visions,
Enabling Technologies, and Challenges,' IEEE Communications Surveys & Tutorials,
vol. 24, no. 2, pp. 1555-1580, 2022. DOI: 10.1109/COMST.2022.3221119.
5. Rahmani et al., 'Computational Resource Allocation in Fog Computing: A
Comprehensive Survey,' ACM Computing Surveys, vol. 56, no. 2, pp. 1-35, 2023. DOI:
10.1145/3586181.
6. S. Yi et al., 'Managing Resources Continuity from the Edge to the Cloud: Architecture
and Performance,' Future Generation Computer Systems, vol. 78, pp. 651-664, 2017.
DOI: 10.1016/J.FUTURE.2017.09.036.

404
 A Comprehensive Review of Resource Management Techniques in Edge Computing

7. B. Sousa et al., 'Resource Aware Placement of IoT Application Modules in Fog-Cloud

Computing Paradigm,' IEEE/IFIP Network Operations and Management Symposium,
pp. 345-352, 2017. DOI: 10.23919/INM.2017.7987464.
8. R. Biswas et al., 'Edge and Fog Computing in Critical Infrastructures: Analysis,
Security Threats, and Research Challenges,' IEEE Euro S & PW, pp. 1-7, 2019. DOI:
10.1109/EUROSPW.2019.00007.
9. Y. Cao et al., 'Resource Management in Distributed Computing,' in Proceedings of the
9th International Conference on Cloud Computing and Services Science, pp. 1-9, 2019.
DOI: 10.1007/978-981-97-2644-8_1.
10. J. Li et al., 'A Systematic Study of Load Balancing Approaches in the Fog Computing
Environment,' Journal of Supercomputing, vol. 77, no. 7, pp. 7389-7410, 2020. DOI:
10.1007/S11227-020-03600-8.

405
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

43. Applications of Machine Learning and Deep

Learning in Farming: A Review
Halaswamy B. M.
Electronics and Communication,
Sri Siddhartha Academy of Higher Education,
Tumkur, India.
Vinutha C. B.
Electronics and Communication,
School of Engineering, Presidency University,
Bengaluru, India.

ABSTRACT:

India is largely an agriculture product producing country. Agriculture in India is facing

many challenges like drought, flood, disease, pricing, yield and etc. Precision agriculture
is one of the rapidly developing fields. To address current challenges in agriculture, Deep
learning stands as a promising technology in precision farming, facilitating the
advancement of sophisticated disease detection and categorization methods. Plant disease
recognition by deep learning, eliminates the need for manual identification of disease
features, rendering feature abstraction more objective and enhancing technological
efficiency. This paper provides a comprehensive review of machine learning and deep
learning techniques applied to detect and classify plant diseases. The paper discusses
available datasets for crop and plant disease detection and abstraction, followed by a
comparative investigation of various algorithms utilized in leaf disease detection.

KEYWORDS:

Classification, Machine learning, Object detection, Deep learning, and Plant leaf disease
detection.

I Introduction:

Indian economic development relies heavily on agriculture, which faces numerous

challenges. Meeting the current food demands of the population has become increasingly
difficult due to factors such as population explosion, uneven weather conditions, and
shortfall of resources. In addition to these challenges, the severity and prevalence of crop
diseases have been on the rise.
406
 Applications of Machine Learning and Deep Learning in Farming: A Review

The presence of plant diseases significantly hampers agricultural production, exacerbating

food insecurity if not addressed promptly. Continuous monitoring is essential to mitigate
production losses caused by crop diseases. Precision agriculture, a rapidly evolving field,
aims to tackle concerns regarding agricultural sustainability. Machine learning (ML) stands
out as a promising technology in this domain, enabling machines to learn autonomously
without direct programming. Numerous studies have leveraged Machine Learning methods
for detection and classification of plant diseases [1][2], primarily utilizing plant or leaf
images as input and categorizing them as healthy or diseased, or performing multiclass
classification for multiple diseases. Machine learning (ML) techniques such as Random
Forest and Deep Learning(DL) have been commonly used for this purpose [1][2]. However,
fewer studies have focused on simultaneously identifying the disease type and the diseased
input image regions [1]. This aspect gains significance in scenarios with multiple plant
diseases or when precise localization of diseased regions in large crop images is required.
Moreover, the object detection challenge poses greater difficulty compared to classification,
and DL methods struggle in uncontrolled environments, such as images with noisy
backgrounds. This paper aims to review the application of ML and DL methods in either
classifying or detecting plant diseases, offering insights into their efficacy within the realm
of precision agriculture. Numerous research papers have explored ML and DL techniques
in the context of precision agriculture, contributing to ongoing efforts to enhance
agricultural productivity and sustainability.

II Literature Review:

This section offers an in-depth examination of disease detection and classification

techniques employing ML and DL methodologies, as evidenced in the existing literature.

A. Disease Classification:

Amara et al. [3] LeNet architecture is used to classify diseases in banana leaves. In their
work they downsized images to 60x60 in the pre-processing steps and converted to
greyscale. This architecture achieved accuracy between 92% to 99% with plant village data
set.

Cruz et al. [4] applied the LeNet architecture to detect indications of olive decline in their
research. They trained the LeNet architecture network using the Plant Village dataset, where
images were pre-processed by resizing them to 256×256. Their reported accuracy reached
an impressive 99%.

De Chant et al. [5] introduced a 3 stage method of employing CNN models to identify
N.L.B. infected maize plants in their investigation. They curetted a bespoke dataset
comprising 1,796 images and achieved an accuracy of 96.7% in their results.

The Lu et al. [6] authors presented a multistage CNN architecture, drawing inspiration from
AlexNet, with the aim of detecting diseases in rice plants. To compile their dataset, they
collected images from both agricultural-pest and insect-pest databases. Before analysis, the
images underwent preprocessing steps, including resizing to 512 × 512 pixels and the
application of the Z.C.A Whitening technique to eliminate data correlation. This model
achieved a notable accuracy of 95.48%.
407
International Journal of Research and Analysis in Science and Engineering

In their study, Oppenheim and Shani [7] harnessed the power of a CNN to categorize
potatoes into five classes. Four categories are infected and one is healthy. Their dataset
includes 400 images of infected potatoes captured through three digital cameras. As part of
the preprocessing phase, they standardized the images by 224×224 pixels and converting
them to grayscale. They achieve the accuracies ranging from 83% when trained with only
10% of the data to an impressive 96% with 90% of the data utilized for training.

Barbedo [8] identified the key factors affecting the design and efficacy of CNNs in the realm
of plant disease identification. PDDB dataset having 50,000 images is used to identify corn
diseases. Additionally, they conducted an investigation into nine factors influencing disease
detection in maize fields. Four different datasets were used for training; achieved the highest
accuracy of 87% with a subdivided dataset.

Lu et al. [9] conducted a study utilizing a high-resolution spectral sensor for the detection
of tomato leaves diseases at various growth stages. They introduced the K-Nearest Neighbor
(KNN) algorithm [1] for identification of the sensor data into 4 groups: early_stage,
late_stage, healthy, and asymptomatic. They used PCA to achieve 85.7% in healthy leaves
and 86.4% in asymptomatic leaves, 73.5% in early stage and 77.1% for late stage leaves.

In their study, Pineda et al. [10] focused on predicting diseases caused by the bacteria
Dickeyadadantii in melon leaves. Using three machine learning (ML) algorithms, namely
LRA, SVM, and ANN. Their investigation demonstrating a superior accuracy of 99.1%,
particularly in classifying images depicting entire leaves.

Al-Saddik et al. [11] examined spectral bands for developing a multispectral camera for
UAV to detect diseased grapevine fields. The targeted disease is highly contagious,
incurable, and can cause significant production losses. To identify the most effective
spectral bands, 2 spectral examination methodologies were employed. One method involved
a feature selection, utilizing the successive projection algorithm. The second method
focused on classic vegetation metrics. SVM and discriminant classifiers were employed in
this study. The accuracy of the models varied depending on the grapevine variety under
consideration. The approach using the successive projection algorithm outperformed the
common vegetation metrics, achieving a classification accuracy exceeding 96%.

Dhingra et al. [12] introduced methodologies aimed at identifying diseases affecting basil
leaves through the application of digital image processing methods. They employed nine
distinct classifiers for this purpose. The process of image acquisition involved gathering
samples from an herb garden, with careful attention given to standardizing the surface
condition of the leaves. The classifiers use Decision Trees (DT), Support Vector Machines
(SVM), linear models, Naive Bayes, K-Nearest Neighbors (KNN), AdaBoost, discriminant
analysis, Random Forests (RF), and Artificial Neural Networks (ANNs). The classification
task aimed to segregate the images into two primary categories such as infected and healthy.
Among the nine classifiers employed, Random Forests (RF) emerged as the most accurate,
achieving an impressive exactness of 98.4%.

Habib et al. [13] introduced agro-medical expert system for the identification of papaya
diseases using computer vision methods to analyze images captured through portable

408
 Applications of Machine Learning and Deep Learning in Farming: A Review

devices. The objective of the study was to disease detection and disease classification. They
achieved a commendable accuracy rate of 90.15%.

Karthik et al. [14] introduced 2 types of Deep Neural Network (DNN) models aimed at
discerning infection present in tomato leaves [1]. One model uses residual learning atop a
feed forward CNN to acquire essential features, and other model incorporated strengthened
attention mechanisms and residual learning within CNNs. These models tested using the
Plant Village dataset, achieving reported accuracies about 95% and 98%.

In their study, Chowdhury et al. [15] investigated the classification of tomato diseases using
the Efficient-Net CNN architecture on tomato leaf images. They employed a dataset
comprising 18,162 tomato images from the Plant-Village dataset for training and fine tuning
the model for detecting both good/healthy and infected tomato leaf images. Impressively,
they achieved a classification accuracy of 0.999 with Efficient_NetB7 and 0.998 with
Efficient_NetB4.

Karlekar and Seal [16] recommended a machine vision technology for identifying and
categorizing leaf diseases within crops of soybean. Their approach involves isolating the
leaf section by removing multifaceted backgrounds from the whole image. Subsequently, a
Convolutional Neural Network (CNN) named SoyNet, trained on the PDDB dataset
comprising 16 classes, classifies the segmented leaf images. During preprocessing, the
images are resized to 100 × 100 pixels. Impressively, this model achieves an identification
accuracy of 0.9814.

B. Disease Detection:

Jiang et al. [17] utilized a deep learning (DL) methodology with GoogLeNet architecture to
detect infections in apple leaves. Custom dataset consisting of 2029 images depicting
unhealthy apple leaves and subsequently trained their algorithm to identify 5 apple leaf
ailments like Alternaria leaf spot, brown spot, mosaic, grey spot, and rust. Their model
achieved a commendable detection accuracy of 0.7880mAP.

Li et al. [18] introduced a methodology for disease identification in rice crops employing a
convolutional neural network. They collected images using mobile phones and used Faster
R-CNN as the underlying framework for image detection. Comparative analysis
demonstrated that their proposed approach outperforms ResNet50 and ResNet101 in terms
of accuracy and efficiency.

Saleem et al. [19] integrated three different algorithms for object detection with four feature
extractors and three optimization techniques to address plant disease identification. Among
these, the S.S.D model linked with the InceptionV2 feature extractor and trained using the
Adam-optimizer achieving a mAP of 73.07%.

Sun et al. [20] introduced a model designed to detect maize-leaf blight infection on maize
crops utilizing the S.S.D. algorithm. They employed a dataset comprising 18,000 images
captured by a camera mounted on a UAV. Their model achieved an impressive accuracy of
91.83%.

409
International Journal of Research and Analysis in Science and Engineering

Xie et al. [21] introduced a deep learning-based detector designed for identifying leaf
infections in grape plants. They used disease dataset consisting of 4,449 original images and
augmented it with an added 62,286 unhealthy leaf images. The model trained using data
augmentation techniques. This approach resulted an accuracy of 81.1%.

Roy and Bhaduri [22] introduced deep learning (DL) model for apple plant disease
detection. They achieved a detection rate of 56.9 frames per second [1] while enhancing the
mean average precision (mAP) up to 91.2% [1].

Selvaraj et al. [23] introduced a pixel-based classification method integrated with machine
learning (ML) models to detect banana crops using multilevel satellite images and
unmanned aerial vehicles (UAVs). They utilized Random Forest (RF) for pixel-based
classification and devised a mixed model strategy incorporating RetinaNet along with
minimized-classifier for banana localization and infection classification [1] using U.A.V
RGB images. Their proposed method produced accuracies of 0.994, 0.928, 0.933, and 0.908
for detecting banana bunchy top disease [1].

III Challenges in Plant Disease Detection:

Following an extensive examination of machine learning (ML) and deep learning (DL)
algorithms for plant/crop disease/infection detection and classification, many challenges in
real field applications of plant disease detection have come to light.

1. Existing models are predominantly on image data, neglecting valuable non-image data
such as environmental data (temperature and humidity). This oversight limits the
understanding of plant health and disease dynamics. Addressing this gap by integrating
nominate data into classification and object detection algorithms [1] is crucial for
enhancing the accuracy and robustness of predictions.
2. The availability of fully annotated open datasets for plant disease research remains
limited. Numerous studies heavily rely on the Plant Village dataset, primarily acquired
under controlled laboratory conditions. There's a pressing need for larger datasets
collected under real-world settings to better reflect diverse environmental conditions
and disease manifestations.
3. While most research approaches views disease detection as a classification problem,
often binary or Multiclass [1], there is growing acknowledgments that object detection
methods can provide more comprehensive insights by not only identifying the type of
disease but also pinpointing the affected regions within the image. Object detection
methodologies have the potential to facilitate more detailed analyses of plant health.
4. Researcher depends on a single dataset for both training and testing their models.
Models trained on single dataset frequently exhibit subpar performance when applied
to different datasets. To improve the model performance, diverse range of datasets could
be used.
5. Instead of relying only on CNNs models, researcher can find neural network
architectures like recurrent- neural- networks (RNN)that improves disease detection.
6. Many researchers use long leaf image data sets for analysis. They can investigate on
small leaves also. That enhances the detection of disease in early stage

410
 Applications of Machine Learning and Deep Learning in Farming: A Review

IV Future Research Scope of Disease Detection in Plants:

Apart from the aforementioned challenges, there exist numerous promising avenues for
future research in the field of plant disease detection.

1. To enhance prediction accuracy in disease detection algorithms, it's imperative to

develop models that can seamlessly integrate nominate data [1], such as environmental
factors. By incorporating additional contextual information beyond visual cues, these
models can provide a more comprehensive understanding of the underlying factors
influencing plant health. This holistic approach enables more accurate and robust
predictions, thereby improving the efficacy of disease detection systems.
2. To bolster the generalizability of models in agricultural settings, it's crucial to
collaborate with domain experts to create diverse and real-world datasets. By capturing
data under varied environmental conditions and agricultural practices, these datasets
can better reflect the complexities of real-world scenarios. This collaborative effort
ensures that models are trained on a representative range of data, improving their ability
to adapt and perform accurately across different agricultural contexts.
3. There's a growing need to prioritize object detection methods in plant disease prediction
endeavors. By leveraging object detection techniques, researchers can glean more
granular insights into disease localization within plant images. This shift in focus
promises to enhance the precision and specificity of disease detection models, enabling
more accurate identification of affected regions and facilitating targeted intervention
strategies.
4. Developed model must be practically implementable.
5. To overcome challenges posed by variable lighting conditions and occluded images, it's
crucial to implement techniques that enhance algorithm robustness. By developing
strategies specifically designed to mitigate the effects of illumination variations and
image occlusions, algorithms can maintain their performance consistency across
diverse environmental settings and image complexities. This proactive approach
ensures that disease detection systems remain reliable and effective under real-world
conditions, ultimately improving their utility in agricultural applications.
6. Efforts to enhance computational efficiency are paramount, which is necessitating a
focus on optimizing model architectures and algorithms to cater to real-time
applications. By streamlining computational processes and reducing resource demands,
optimized models can deliver swift and responsive performance, crucial for deployment
in dynamic agricultural settings. This optimization drive ensures that disease detection
systems not only maintain high accuracy but also operate seamlessly in real-world
scenarios, maximizing their practical utility and impact.

After the extensive investigation of the different existing work in this area, a brief outline
of the research gap, reasons, and probable solutions are being demonstrated in the Table I.

411
International Journal of Research and Analysis in Science and Engineering

Table I: Outline of The Research Gaps, Reasons and Solutions of Some of Previous
Works

Source Crop Dataset Algorithms Research Gap Reason Solutions

Banana Plant LeNet Fixed/predefined Difficulty in Getting real
Village data set getting real world data
world data. using UAV
This leads no and
3
practical classifying
applications. diseases is
useful for
Agriculture.
Olive Plant CNN Time consuming They trained Models
Symptoms Village the model need to be
4 with two improved.
different
data set.
Maiz- Custom CNN Three stage Data set is Size of the
plants dataset approach used only 1796 dataset need
5
images. to be
Improved.
Rice Plants Own CNN Own database Many Real time
Database with less images images were data is
taken from required for
6
book. achieving
good
accuracy.
Potatoes Own data CNN Own database 400 Images Size of the
Base with less images were taken dataset need
from simple to be
digital increased.
7
camera.
Resolution
was the
problem.
Corn PDDB CNN Achieved Multiple Experiment
accuracy is only data set with with single
8 87% many dataset with
subdivision. real time
Images.
Tomato Custom KNN Accuracy is less Data set is Data Set and
about 73.5% not sufficient algorithms
9
for analysis. need to be
improved.

412
 Applications of Machine Learning and Deep Learning in Farming: A Review

Source Crop Dataset Algorithms Research Gap Reason Solutions

Tomato Plant DNN Time consuming. Used two Select
14 Leaves Village models. Different
models.
Tomato Plant Efficient Two Numbers of Improve the
Leaf Village Net CNN segmentation parameters models
15 images models used are more. using
minimum
parameters.

V Conclusion:

This study gave us insight into existing research employed utilizing machine learning (ML)
and deep learning (DL) methods used for farming. This study was highlighting on
methodologies for plant and crop disease detection and classification. Here, introduced a
classification scheme that classifies related works into distinct classes. These studies are
divided into two categories based on methodology, namely classification and object
detection approaches. The review here presented an overview of available datasets for plant
disease detection and classification, offering insights into their respective classes, data
characteristics, and suitability for either classification or object detection tasks. This
comprehensive analysis aims to provide valuable insights and guidance for researchers in
farming.

References:

1. Vasileios Balafas, Emmanouil Karantoumanis, Malamati Louta, Nikolaos Ploskas.

"Machine Learning and Deep Learning for Plant Disease Classification and Detection",
IEEE Access, vol. 11, pp. 114352 -114377, 2023.
2. Ruchi Rani, Jayakrushnaahoo, Sivaiah Bellamkonda, Sumit Kumar, Sanjeev Kumar
Pippal. "Role of Artificial Intelligence in Agriculture: An Analysis and Advancements
with Focus on Plant Diseases", IEEE Access, vol. 11, pp. 137999-138019
3. J. Amara, B. Bouaziz, and A. Algergawy, ‘‘A deep learning-based approach for banana
leaf diseases classification,’’ in Proc. Datenbanksys.Für Bus., Technol. Web (BTW),
Workshop band, CA, USA, Jul. 2017, pp. 1–24.
4. A. C. Cruz, A. Luvisi, L. De Bellis, and Y. Ampatzidis, ‘‘X-FIDO: An effective
application for detecting olive quick decline syndrome with deep learning and data
fusion,’’ Frontiers Plant Sci., vol. 8, p. 1741, Oct. 2017.
5. C. DeChant, T. Wiesner-Hanks, S. Chen, E. L. Stewart, J. Yosinski, M. A. Gore, R. J.
Nelson, and H. Lipson, ‘‘Automated identification of northern leaf blight-infected
maize plants from field imagery using deep learning,’’ Phytopathology, vol. 107, no.
11, pp. 1426–1432, Nov. 2017.
6. Y. Lu, S. Yi, N. Zeng, Y. Liu, and Y. Zhang, ‘‘Identification of rice diseases using deep
convolutional neural networks,’’ Neuro computing, vol. 267, pp. 378–384, Dec. 2017.
7. Oppenheim and G. Shani, ‘‘Potato disease classification usingconvolution neural
networks,’’ Adv. Animal Biosci., vol. 8, no. 2, pp. 244–249, 2017.
8. J. G. A. Barbedo, ‘‘Factors influencing the use of deep learning for plant disease
recognition,’’ Biosyst. Eng., vol. 172, pp. 84–91, Aug. 2018.
413
International Journal of Research and Analysis in Science and Engineering

9. J. Lu, R. Ehsani, Y. Shi, A. I. de Castro, and S. Wang, ‘‘Detection of multi tomato leaf
diseases (late blight, target and bacterial spots) in different stages by using a spectral-
based sensor,’’ Sci. Rep., vol. 8, no. 1, pp. 1–11, Feb. 2018.
10. M. Pineda, M. L. Pérez-Bueno, and M. Barón, ‘‘Detection of bacterial infection in
melon plants by classification methods based on imaging data,’’ Frontiers Plant Sci.,
vol. 9, p. 164, Feb. 2018.
11. Al-Saddik, J. C. Simon, and F. Cointault, ‘‘Assessment of the optimal spectral bands
for designing a sensor for vineyard disease detection: The case of ‘Flavescencedorée,’’’
Precis. Agricult., vol. 20, no. 2, pp. 398–422, Apr. 2019.
12. Dhingra, V. Kumar, and H. D. Joshi, ‘‘A novel computer vision based neutrosophic
approach for leaf disease identification and classification,’’ Measurement, vol. 135, pp.
782–794, Mar. 2019.
13. M. T. Habib, A. Majumder, A. Z. M. Jakaria, M. Akter, M. S. Uddin, and F. Ahmed,
‘‘Machine vision based papaya disease recognition,’’ J. King Saud Univ.-Comput. Inf.
Sci., vol. 32, no. 3, pp. 300–309, Mar. 2020.
14. R. Karthik, M. Hariharan, S. Anand, P. Mathikshara, A. Johnson, and R. Menaka,
‘‘Attention embedded residual CNN for disease detection in tomato leaves,’’ Appl. Soft
Comput., vol. 86, Jan. 2020, Art. no. 105933.
15. M. E. H. Chowdhury, T. Rahman, A. Khandakar, M. A. Ayari, A.U. Khan, M. S. Khan,
N. Al-Emadi, M. B. I. Reaz, M. T. Islam, and S. H. M. Ali, ‘‘Automatic and reliable
leaf disease detection using deep learning techniques,’’ Agri Engineering, vol. 3, no. 2,
pp. 294–312, May 2021.
16. A. Karlekar and A. Seal, ‘‘SoyNet: Soybean leaf diseases classification,’’ Comput.
Electron. Agricult., vol. 172, May 2020, Art. no. 105342.
17. P. Jiang, Y. Chen, B. Liu, D. He, and C. Liang, ‘‘Real-time detection of apple leaf
diseases using deep learning approach based on improved convolutional neural
networks,’’ IEEE Access, vol. 7, pp. 59069–59080,2019.
18. Li, R. Wang, C. Xie, L. Liu, J. Zhang, R. Li, F. Wang, M. Zhou, and W. Liu, ‘‘A
recognition method for rice plant diseases and pests video detection based on deep
convolutional neural network,’’ Sensors, vol. 20, no. 3, p. 578, Jan. 2020.
19. M. H. Saleem, S. Khanchi, J. Potgieter, and K. M. Arif, ‘‘Image-based plant disease
identification by deep learning meta-architectures,’’ Plants, vol. 9, no. 11, p. 1451, Oct.
2020.
20. Sun, Y. Yang, X. He, and X. Wu, ‘‘Northern maize leaf blight detection under complex
field environment based on deep learning,’’ IEEE Access, vol. 8, pp. 33679–33688,
2020.
21. X. Xie, Y. Ma, B. Liu, J. He, S. Li, and H. Wang, ‘‘A deep-learning-based real-time
detector for grape leaf diseases using improved convolution nalneural networks,’’
Frontiers Plant Sci., vol. 11, p. 751, Jun. 2020.
22. A. M. Roy and J. Bhaduri, ‘‘A deep learning enabled multi-class plant disease detection
model based on computer vision,’’ AI, vol. 2, no. 3, pp. 413–428, Aug. 2021.
23. M. G. Selvaraj, A. Vergara, F. Montenegro, H. A. Ruiz, N. Safari, D. Raymaekers, W.
Ocimati, J. Ntamwira, L. Tits, A. B. Omondi, andG. Blomme, ‘‘Detection of banana
plants and their major diseases through aerial images and machine learning methods: A
case study in DR Congo and republic of Benin,’’ ISPRS J. Photogram. Remote Sens.,
vol. 169, pp. 110–124, Nov. 2020.
24. M. Brahimi, K. Boukhalfa, and A. Moussaoui, ‘‘Deep learning for tomato diseases:
Classification and symptoms visualization,’’ Appl. Artif. Intell., vol. 31, no. 4, pp. 299–
315, Apr. 2017.
414
 Applications of Machine Learning and Deep Learning in Farming: A Review

25. Abdulridha, R. Ehsani, A. Abd-Elrahman, and Y. Ampatzidis, ‘A remote sensing

technique for detecting laurel wilt disease in avocado in presence of other biotic and
abiotic stresses,’’ Comput. Electron. Agricult., vol. 156, pp. 549–557, Jan. 2019.
26. Y. Zhang, C. Song, and D. Zhang, ‘‘Deep learning-based object detection improvement
for tomato disease,’’ IEEE Access, vol. 8, pp. 56607–56614,2020.

415
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

44. Aggregation Based on Clustered Data (ABCD)

with Edge Computing
Achyut Yaragal, M. N. Deshmukh
Research Scholar,
Department of Electronics & Communication
Basaveshwara Engineering College,
Bagalkote, Karnataka.
Kirankumar Bendigeri, S. B. Kumbalavati
Assistant Professor,
Department of Electronics & Communication,
Basaveshwara Engineering College,
Affiliated to V.T.U, Belagavi,
Bagalkote, Karnataka, India.
J. D. Mallapur
Professor,
Department of Electronics & Communication,
Basaveshwara Engineering College,
Affiliated to V.T.U, Belagavi, Bagalkote,
Karnataka, India.

ABSTRACT:

In the era of big data and the Internet of Things (IoT), efficient data processing is crucial.
This paper explores the integration of cluster-based data aggregation with edge computing
and machine learning to enhance data processing efficiency, reduce latency and improve
decision-making in real-time applications. We present a comprehensive framework to
discuss key benefits and challenges, and illustrate the implementation with a case study in
a smart city environment.

KEYWORDS:

Data Aggregation, Clustered Data Edge Computing, Edge Devices.

416
 Aggregation Based on Clustered Data (ABCD) with Edge Computing

I Introduction:

A wireless network, or WSN, is made up of several nodes and base stations. These networks
work together to cooperatively transfer data to the main location while keeping an eye on
environmental or physical parameters like temperature, sound, pressure, and so on.

Sensor nodes in the Internet of Things form wireless sensor networks (WSN) crucial for
various applications, but face challenges due to limited energy resources.

High energy consumption poses a significant challenge for sensor nodes, leading to
shortened network life cycles.

The non-rechargeable and restricted power supply of nodes has prompted research into
novel approaches to enhance the energy balance and energy efficiency of wireless sensor
networks.

The different challenges in wireless sensor networks include the following.

• Fault Performance: Physical harm could happen because some sensor nodes stop
functioning due to power outages. The ability of a sensor network to continue operating
normally even in the event that a sensor node fails is known as fault tolerance.
• Scalability: Routing systems should be scalable enough to handle events, and the
number of nodes employed in the detecting region could be in the range of hundreds to
thousands.
• Quality of Service: The application may require certain levels of quality of service, such
as energy efficiency, long lifetime, and trustworthy data.
• Data Aggregation: Data aggregation is the process of combining data from multiple
sources with various functions, such as average, max, and min.
• Data Compression: Procedure for altering data to make it smaller Data compression
can reduce storage requirements, expedite file transfers, and lower expenses associated
with storage hardware and network bandwidth.
• Data Latency: Data latency is the total time elapsed between when data are acquired
by a sensor and when these data are made available to the public.

Edge Devices: Positioned between end devices and the central cloud, responsible for local
data processing, analysis, and communication management. They are more powerful and
reduce the load on central systems.

End Devices: Simple devices at the edge of the network that collect data or perform actions
based on received instructions. They rely on edge devices or central systems for complex
processing tasks.

II. Literature Survey/Related Works.

A sensor is a device/apparatus that can react to and identify input from environmental or
physical factors, such as pressure, heat, light, etc. An electrical signal that is sent to a
controller for additional processing is the sensor's output.
417
International Journal of Research and Analysis in Science and Engineering

Proposes a two-layer WSN system based on edge computing to address challenges related
to high energy consumption and short life cycle of WSN data transmission [1]. Establishes
optimization models for wireless energy consumption and distance in sensor networks.
Sparrow search algorithm is used to evenly distribute sensor nodes in the system, reducing
resource consumption and prolonging the network's life cycle.

An efficient data aggregation scheme (EDAS) for IoT-based wireless sensor networks
(WSN) to address issues like data redundancy, high computation overhead, and high energy
consumption. It utilizes an improved low energy adaptive clustering algorithm (I-LEACH)
to form an optimal number of cluster heads (CH) based on node residual energy and average
network energy. Network coding is used to eliminate data redundancy through linear XOR
operations, ensuring non-replicated data transmissions [2].

Vehicle clustering has received attention for a promising approach to enhancing in network
stability, reliability, and scalability [3]. An efficient greedy algorithm is developed, which
by considering vehicle states over a prediction horizon, identifies a set of cluster heads and
their designated cluster members. We leverage the capabilities of Vehicular Edge
Computing servers and implement a load-balancing approach that distributes computation
tasks among the VEC servers to prevent the overloading of any single server.

A data aggregation approach based on Adaptive Huffman Coding, which by identifying and
eliminating duplicate data reduces the volume of clustering data transmitted to VEC servers.

One of the main challenges in constructing wireless sensor networks (WSNs) is the limited
energy of the sensor nodes. One useful tactic to reduce energy usage and increase network
lifespan is data aggregation [4]. The best method has been shown to be cluster-based data
aggregation due to its benefits, which include low computing overhead, flexible, scalable,
accurate, reliable, and accurate data processing.

The Internet of Things (IoT) and machine learning (ML) have developed quickly, and large
amounts of data produced by edge devices like laptops, smartphones, and artificial
intelligence (AI) speakers have been utilized extensively to train ML models [5]. In Multi-
Access Edge Computing, we employed the blockchain technique based on clusters.

Sensor networks use clustering strategies to minimize communication overheads, guarantee

optimal resource utilization, lower total system energy consumption, and minimize inter-
SN interference.

The primary purpose of clustering routing is to lower the data transmission rate by using
the Cluster Head's (CH) information pooling technique.

The three energy-intensive functions of SNs are sensing, processing, and communication.
According to technical standards, the processor needs the same amount of energy to
transport one bit of data as it does to execute several arithmetic operations. Furthermore,
practically all SNs can provide a similar data rate due to the physical environment of a
heavily deployed SN network, meaning that transmitting such data is redundant. Therefore,
it is essential to combine all the elements that promote SN clustering in a way that makes it
possible to transfer only compact data.
418
 Aggregation Based on Clustered Data (ABCD) with Edge Computing

The Clustering Algorithms challenges [6]

Determining the optimal number of clusters: Selecting the right number of clusters for the
data is one of the main issues in clustering. Improper clustering outcomes can arise from
selecting an inappropriate number of clusters.

Handling high: dimensional data: The ills of dimensionality mean that clustering
techniques may perform poorly on high-dimensional data. The significance of the distance
or similarity metrics between data points decreases with the number of dimensions.
Additionally, low density problems and increased computational cost in clustering
techniques can be caused by high-dimensional data.

Sensitivity to initialization: Numerous clustering techniques are dependent on the initial

configuration, including K-means. It might be difficult to get consistent and trustworthy
findings from clustering since different initializations can produce varied results.

Dealing with different cluster shapes and sizes: Convex or isotropic clusters are two
examples of the underlying structures that are commonly assumed by clustering methods.
However, clusters with irregular shapes, variable sizes, or overlapping boundaries are
frequently found in real-world data.

Handling noisy or outlier data: Due to their ability to distort cluster boundaries and create
misleading clusters, noisy data points can have a substantial impact on the clustering
process.

Scalability to large datasets: Some clustering techniques have memory or processing

complexity issues that make them difficult to scale to huge datasets. The clustering method
may grow more time-consuming or perhaps impossible as the dataset grows.

To overcome these obstacles, one must carefully analyze the dataset's properties, choose the
best algorithms, and use preprocessing or algorithmic changes.

To guarantee the caliber and dependability of the clustering results, it is crucial to assess
and validate the clustering results using domain expertise and extra studies.

Managing the sensory data is a tedious task in sensor cloud. Usually sensor nodes produce
multiple data and have heterogeneity character. Fog computing is a new paradigm to remove
the latency problem and improves the system accuracy. Fog computing is a middleware
between end devices and cloud server.

As number of users increase, the resource allocation becomes very difficult in sensor cloud.
In this paper [7], they have proposed a methodology for resource provision and pricing
model for sensor cloud. The resource allocation is completed on priority basis as requested
by the user.

419
International Journal of Research and Analysis in Science and Engineering

Vehicular Ad hoc Networks mainly depends on the cloud computing for the services like
storage, computing and networking. With the increase in the number of vehicles connected
to the cloud, the problems like network congestion and increased delay arises. Hence fog
enhanced vehicular services having several advantages compared to the cloud-only model.
Resources at the fog layer are less compared to the cloud but with the proper utilization of
these resources, it is possible to serve the requesting vehicles diverted towards it [8].

Heterogeneous Sensor Nodes: Refers to a collection of sensor nodes that differ in terms of
capabilities, types of sensors, communication protocols, power consumption, and
computational power. It also refers to the diversity within the network in terms of hardware,
software & functionality.

These nodes often communicate with each other within a localized network (e.g., WSN)
and may require gateways or aggregators to connect to external networks or the internet.

The focus is on localized data collection, possibly with on-site data processing or
aggregation before transmission to a central server. Typically deployed in specific
environments like industrial monitoring, environmental sensing, or military applications.
Hence heterogeneous sensor nodes refer to a collection of diverse sensors within a network.

IoT Sensor Devices: They are designed to collect data and communicate it over the internet,
often integrating with cloud platforms for processing and analysis. They are generally
designed to connect to the internet directly or through a gateway.

Generally designed to be low-power and efficient, with some capable of energy harvesting.
These devices may have more integrated processing power and memory to handle data
preprocessing, reducing the need for constant communication with external servers.

They are often part of a larger ecosystem and data is not only collected but also used for
real-time decision-making, automation, and remote monitoring.

IoT devices may also support edge computing, where some processing is done locally
before data is sent to the cloud, enabling faster response times and reduced bandwidth usage

III. Proposed Scheme for System Architecture:

Our proposed framework integrates cluster-based data aggregation, edge computing, and
machine learning. The architecture consists of three main layers: data collection, edge
processing, and centralized analysis. The Flow chart of the main three layers is being
separated into many layers as shown in the Figure 1. This flow chart initial Phase consists
of data collection in which many sensors will send the environment signal to the processor.
Once sensor network is formed and them after the clustering process will take place, once
clustering is done cluster head is selected within the group which node is having more
energy.

Data collected need to be filter / preprocessed for further work after that the data will have
computation.

420
 Aggregation Based on Clustered Data (ABCD) with Edge Computing

Figure 1: Flow Chart of the Data Aggregation in Clustering Methods with Edge
Computing

A. Data Collection:

IoT devices collect raw data, which is grouped into clusters based on predefined criteria,
such as geographical location or data type. Data is collected from various IoT devices
distributed across different locations. These devices gather raw data on parameters such as
environmental conditions, traffic, and energy consumption. Data is initially clustered based
on criteria like geographical proximity, data type, and temporal aspects.

B. Cluster-Based Data Aggregation:

Since the CHs serve as the entry point between the SNs and the BS, choosing a cluster head
comes first in most clustering schemes. Since the CH serves as a communication mediator
between the BS and the SNs, choosing the CH is an important step in the subsequent
clustering processes that increase the network's lifespan and energy efficiency.
421
International Journal of Research and Analysis in Science and Engineering

Clustering Characteristics:

• Inter-cluster head connectivity: Indicates how well SNs or CHs are able to
communicate with the BS. In the event that the CH is unable to establish long-distance
communication, the clustering scheme must offer the BS intermediate routing paths.
• Cluster count: It is a measure of how many clusters are created in a given round; the
more CHs, the smaller the cluster distribution size and the higher the level of energy
conservation. Certain clustering techniques involve pre-assigned CH selection, which
allows the CHs to be chosen at random and produces varying amounts of clusters.
• Cluster size: The ideal path length in a cluster between each node and its distance from
the CH. Energy consumption improves with decreasing cluster size since less
transmission distance and CH load are required. Certain clustering techniques have a
fixed cluster size, particularly when the clusters are fixed for the duration of their service
life; however, other clustering techniques have a variable cluster size.
• Cluster density: It refers to the quantity of regular nodes inside a cluster, it is a laborious
work for the CHs in dense clusters to reduce energy usage. Because of this, the majority
of clustering techniques use sparse cluster density and fixed clustering.
• Message count: It is used to describe the minimum amount of message transmissions
needed to pick a CH. The CH selection process requires a greater amount of energy the
more messages there are. The majority of non-probabilistic algorithms need message
transfer in order to choose the CH.
• Stability: If a cluster's members are not fixed, then clustering methods are adaptable if
not, they are fixed since the cluster count cannot be changed during the CH selection
process. Enhancing the cluster count increases an SN's stability.

Clustering Methods:

• K-Means: Groups data into K clusters by minimizing within-cluster variance.

• Hierarchical Clustering: Creates a dendrogram to represent data groupings.
• DBSCAN: Clusters data based on density, identifying outliers effectively.

Aggregation Techniques:

• Summarization: Calculating statistical summaries like mean, median, and mode.

• Sampling: Selecting representative data points from each cluster.
• Compression: Reducing the data size using compression algorithms.

C. Edge Processing:

Edge nodes preprocess the clustered data, performing tasks such as filtering, noise
reduction, and preliminary analysis. Simple machine learning models deployed at the edge
detect anomalies and make quick decisions.

1. Preprocessing:

• Filtering: Removing noise and irrelevant data.

• Normalization: Standardizing data to a common scale.
422
 Aggregation Based on Clustered Data (ABCD) with Edge Computing

• Feature Extraction: Identifying and extracting important features from the raw data.

2. Local Analysis:

• Anomaly Detection: Using lightweight machine learning models to detect unusual

patterns.
• Real-Time Analytics: Performing quick, on-the-fly analysis to support immediate
decision-making.

3. Communication:

• Data Transmission: Transmitting processed data to central servers for further analysis.
• Decentralized Coordination: Edge nodes coordinate with each other to ensure
consistency and avoid redundant processing.

Complex machine learning models run on central servers for in-depth analysis. These
models aggregate insights from multiple edge nodes to provide comprehensive analytics
and long-term predictions.

The term “machine learning” was coined by Arthur Samuel, and the models play a
prominent role in our daily lives, which is a branch of artificial intelligence (AI) and
computer science that focuses on the using data and algorithms to enable AI to imitate the
way that humans learn, gradually improving its accuracy i.e. is to figure out how we can
build process/systems that improve over time and with repeated use.

Edge devices often require lightweight machine learning models and methods due to their
limited computational resources, power constraints, and the need for real-time processing.
Here are some popular lightweight machine learning models and methods commonly used
for edge devices:

1. Tiny Machine Learning: Is a growing field focused on developing machine learning

models that are small enough to run on microcontrollers and other edge devices with limited
resources.

The benefits of Tiny ML

• Latency: The data does not need to be transferred to a server for inference because the
model operates on edge devices. Data transfers typically take time, which causes a slight
delay. Removing this requirement decreases latency.
• Energy savings: Microcontrollers need a very small amount of power, which enables
them to operate for long periods without needing to be charged. On top of that, extensive
server infrastructure is not required as no information transfer occurs: the result is
energy, resource, and cost savings.
• Reduced bandwidth: Little to no internet connectivity is required for inference. There
are on-device sensors that capture data and process it on the device. This means there
is no raw sensor data constantly being delivered to the server.

423
International Journal of Research and Analysis in Science and Engineering

• Data privacy: Your data is not kept on servers because the model runs on the edge. No
transfer of information to servers increases the guarantee of data privacy.

2. Quantized Models: Quantization reduces the precision of the numbers used in a model
(e.g., from 32-bit floating-point to 8-bit integers), which reduces the model size and
computational requirements.

3. AutoML for Edge Devices: AutoML tools automate the design of machine learning
models, optimizing them for deployment on edge devices by focusing on model size,
latency, and accuracy. Customized machine learning models for specific tasks like object
detection, where edge deployment is crucial. AutoML Vision Edge allows you to train and
deploy low-latency, high accuracy models optimized for edge devices. AutoML Tables
enables your entire team to automatically build and deploy state-of-the-art machine learning
models on structured data at massively increased speed and scale.

IV. Results and Discussion:

• Efficiency: As an alternative to depending on data centers located hundreds of miles

away, data can be processed and stored closer to the devices. As a result, network
transport energy usage may be significantly reduced, and edge computing's low latency
may be advantageous. Lightweight Algorithms versions suitable for resource-
constrained edge devices. Implementing approximate clustering algorithms that trade
off a bit of accuracy for significant gains in speed and reduced resource usage. Develop
incremental or online clustering methods that can update clusters as new data arrives
without requiring a full re-computation.
• Real-Time Analysis: Edge Computing Optimization Distribute the clustering tasks
across multiple edge nodes to balance the computational load which dynamically assign
tasks based on current node workloads and network conditions.

Implement a multi-tier edge architecture where initial clustering is done at the most
peripheral edge nodes (e.g., sensors or gateways), and more complex processing is done at
higher-level edge servers. This reduces the data volume transmitted upstream and leverages
local processing capabilities.

Optimize resource allocation on edge nodes by prioritizing tasks and efficiently managing
CPU, memory, and storage.

Enable direct communication between edge nodes to share intermediate results and
collaborate on clustering tasks, reducing the dependency on central servers and minimizing
latency.

Minimize the amount of data transmitted between edge nodes and central servers by
transmitting only essential summaries or compressed data rather than raw data. There are
several challenges of integrating edge computing with machine learning requires
specialized expertise. Resource constraints on edge devices may limit the deployment of
sophisticated models. Future work will focus on optimizing resource allocation and
enhancing model deployment strategies.

424
 Aggregation Based on Clustered Data (ABCD) with Edge Computing

V. Conclusion:

Integrating cluster-based data aggregation with edge computing offers a powerful approach
to efficient, real-time data processing. This framework demonstrates significant
improvements in latency reduction, data efficiency, and decision-making capabilities.
Continued research and development in this area will further enhance the potential of IoT
and edge computing data applications.

References:

1. Shaoming, Qiu., Jiancheng, Zhao., X., Zhang., Ao, Li., Yahui, Wang., Fen, Chen.
"Cluster Head Selection Method for Edge Computing WSN Based on Improved
Sparrow Search Algorithm." Sensors, 23 (2023).:7572-7572. doi: 10.3390/s23177572s
2. Vani, S, Badiger., T.S, Ganashree. (2022). Data aggregation scheme for IOT based
wireless sensor network through optimal clustering method. Measurement: Sensors,
24:100538-100538. doi: 10.1016/j.measen.2022.100538
3. Ali, Jalooli., Frankie, Murcia. "Vehicular Edge Computing-Driven Optimized Multihop
Clustering with Data Aggregation." (2023).:135-143. doi:
10.1109/cloudnet59005.2023.10490017
4. Aparna, Shinde., R., S., Bichkar. "Energy Efficient Cluster Based Secured Data
Aggregation Using Genetic Algorithm for WSN." (2023).:1-6. doi:
10.1109/asiancon58793.2023.10270293
5. Chih-Peng, Lin., Hui, Yu, Fan. "Cluster-Based Blockchain Systems for Multi-access
Edge Computing." (2024).:103-114. doi: 10.1007/978-981-99-9342-0_12
6. Jubair, A.M.; Hassan, R.; Aman, A.H.M.; Sallehudin, H.; Al-Mekhlafi, Z.G.;
Mohammed, B.A.; Alsaffar, M.S. Optimization of Clustering in Wireless Sensor
Networks: Techniques and Protocols. Appl. Sci. 2021, 11, 11448.
https://doi.org/10.3390/app112311448
7. Sangulagi, P. and Sutagundar, A. (2019). Agent based Dynamic Resource Allocation in
Sensor Cloud using Fog Computing. International Journal on Emerging Technologies,
10(2): 122-128.
8. Hatti, D.I., Sutagundar, A.V. (2020). Agent Technology Based Resource Allocation for
Fog Enhanced Vehicular Services. In: Balaji, S., Rocha, Á., Chung, YN. (eds)
Intelligent Communication Technologies and Virtual Mobile Networks. ICICV 2019.
Lecture Notes on Data Engineering and Communications Technologies, vol 33.
Springer, Cham. https://doi.org/10.1007/978-3-030-28364-3_8

425
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

45. Challenges and Future Prospects of Thin Film

Deposition Techniques: A Critical Review
Vinay Shettar, Sneha B. Kotin,
B. G. Sheeparamatti
Department of Electronics and Communication Engineering
Biluru Gurubasav Mahaswamiji Institute of Technology,
Mudhol.
Manjula Sutagundar
Department of Electrical and Electronics Engineering
Basaveshwar Engineering College, Bagalkot.

ABSTRACT:

Thin film technology is widely utilized across multiple industries to enhance the mechanical,
physical, and chemical properties of bulk materials. This paper provides an overview of the
current status, challenges, and future potential of thin film deposition techniques. It delves
into various thin film deposition processes, detailing their distinct characteristics, future
prospects, and common applications - such as enhancing energy efficiency, wear resistance,
and corrosion resistance. The primary focus is on physical and chemical vapor deposition
techniques. Generally, thin films with minimal thickness are produced through physical
vapor deposition (PVD) and chemical vapor deposition (CVD). While PVD has its
limitations, it remains a valuable method and is often more advantageous than CVD for
depositing thin film materials. The paper also explores notable similarities and differences
between these specific methods, and categorizes sub-methods that share common
principles.

KEYWORDS:

Thin Film, deposition methods, evaporation, Sputtering.

I. Introduction:

Thin film deposition refers to the process of applying thin film coatings onto a substrate
material, which can be composed of various materials and possess a diverse range of
characteristics capable of enhancing or altering the substrate's performance [1]. Because of
their potential technological value and scientific interest in their properties, a wide variety
of materials have been synthesized as thin films.
426
 Challenges and Future Prospects of Thin Film Deposition Techniques: A Critical Review

These thin films serve a broad range of applications, from nanoscale structures to large
coatings on window glass. Researchers have explored numerous techniques in order to
identify the most reliable and cost-effective methods for producing thin films. These
techniques are generally categorized into two main groups based on the deposition process:
physical deposition and chemical deposition methods [2].

The structure of the review work is as follows: Section 2 offers a concise comparison
between Chemical Vapor Deposition and Physical Vapor Deposition. Section 3 outlines the
various methods used for the deposition of thin films.

Figure 1: Different Thin Film Deposition Techniques.

II. Comparative study of PVD and CVD:

The deposition process is categorized into two types: physical vapor deposition (PVD) and
chemical vapor deposition (CVD), based on the distinct principles underlying the film
deposition methods [3]. The contrast between PVD and CVD is detailed in the following
table. The primary disparity between PVD and CVD lies in the physical state of the coating
material: in PVD, the material is in solid form, whereas in CVD, it is in gaseous form.

427
International Journal of Research and Analysis in Science and Engineering

Table I: Difference Between PVD and CVD

Description CVD PVD

Source material Gas precursor solid
state
Process Chemical reaction in gaseous Evaporation and collision
phase, importance of fluid impact on the solid target
dynamics
Temperatures Higher, appx 600-1100 °C Lower, appx below 450 °C
Layers Multi-layer possible Majorly single layer
Coating thickness Thicker and thin both Thin
Wear resistance, High, low Less, higher
toughness
State of stress in Tensile Compressive
coating
Used for Elements, compounds, alloys Alloys easily
difficult
Process, Material Each process is material specific One process, many materials
Line of sight Not needed Needed
Step Coverage Better (50-100%), can even fill Poor
gaps

A. Physical Vapour Deposition (PVD) Method:

PVD, which stands for Physical Vapor Deposition, comprises a range of thin film deposition
techniques. Within PVD, a solid material is vaporized within a vacuum environment and
then applied onto substrates either as a pure material or an alloy composition. This
vaporization process either causes the material to evaporate or to be sputtered, generating a
gaseous plume or beam that deposits a film onto the substrate. PVD refers to a collection of
thin film deposition techniques in which a solid material is vaporized within a vacuum and
then deposited onto a substrate [4].

Coatings generated through this technique demonstrate high durability, resistance to

scratching and corrosion, and suitability for a variety of applications, from solar cells to
eyeglasses and semiconductors. PVD presents numerous advantages, such as creating tough
coatings that can resist corrosion and scratching, as well as producing thin films capable of
withstanding high temperatures. However, PVD may involve higher costs compared to
other thin film deposition methods, and the expenses can vary across different PVD
techniques. Furthermore, PVD is an environmentally friendly process, as it reduces the use,
management, and disposal of toxic substances in comparison to "wet" processes involving
fluid precursors and chemical reactions. Due to its capacity to generate exceptionally pure,
clean, and durable coatings, Physical Vapor Deposition is particularly favored in the
surgical and medical implant industries. The most prevalent types of PVD include
evaporation and magnetron sputtering [5].

428
 Challenges and Future Prospects of Thin Film Deposition Techniques: A Critical Review

Figure 2: Different Physical Vapour Deposition Methods.

III. Deposition Methods:

In the evaporation process, the thin film material—such as metals, alloys, or compounds—
is heated until it reaches its evaporation point and transforms into vapor. This vapor then
moves through the vacuum chamber and deposits onto the substrate, creating a thin layer.
The substrate is typically positioned at a specific angle or distance from the evaporation
source to manage the thickness and uniformity of the deposited film.

The main factors influencing the evaporation process are the temperature of the evaporation
source, the pressure within the vacuum chamber, the distance between the source and the
substrate, and the angle of the substrate[6-7]. By adjusting these parameters, the desired
properties of the thin film—such as thickness, composition, and surface morphology—can
be achieved.

In essence, the evaporation process utilized for thin film deposition entails heating a
material to the point of vaporization, after which the vapor is deposited onto a substrate to
create a thin layer. This method is extensively utilized due to its capability to generate
consistent, high-quality thin films ideal for a diverse array of applications, such as optics,
electronics, and coatings [8].

A. Electron-beam evaporation:

In E-beam evaporation, a vacuum chamber is utilized along with an electron gun, a material
source (typically in a crucible or filament form), and a substrate holder. The vacuum
environment is crucial to prevent contamination and unwanted reactions during deposition.

The process starts with the material to be deposited placed in the crucible or filament inside
the vacuum chamber. Once the vacuum is established, the electron gun emits a high-energy
electron beam directed at the material source. This bombardment of electrons causes rapid
heating of the material, leading to its sublimation from solid to vapor phase.

429
International Journal of Research and Analysis in Science and Engineering

The material that has been vaporized travels in straight paths within the vacuum chamber
and then condenses onto an adjacent substrate, usually cooled to guarantee uniform film
growth and property control. E-beam evaporation enables precise management of
deposition rates and film thickness, facilitating the creation of top-quality films with
exceptional adhesion and uniformity. This method is capable of depositing a wide variety
of materials, encompassing metals, oxides, and semiconductors [9].

Nevertheless, E-beam evaporation has its constraints. It can potentially harm delicate
substrates as a result of high-energy electron bombardment, and certain materials may
demonstrate inadequate thermal stability or reactivity with the electron beam. Furthermore,
the procedure may be relatively sluggish when weighed against alternative deposition
methods like sputtering. Despite these limitations, E-beam evaporation continues to be a
versatile and essential technique for thin film deposition, playing a substantial role in
driving progress across various technological applications.

B. Molecular Beam Epitaxy (MBE):

Molecular Beam Epitaxy (MBE) stands as an advanced method employed for depositing
thin films with precision at the atomic level. Below is an elaborate overview of MBE and
its applications:

1) What is MBE?

MBE is a method utilized for precision deposition of thin films at the atomic level. This
technique entails the deposition of individual atoms or molecules onto a substrate, leading
to the creation of high-quality crystalline films.

2) Process of MBE:

• Evaporation: In MBE, the typical process entails heating solid source materials within
a vacuum to produce a stream of evaporated atoms or molecules. Subsequently, these
evaporated species are directed onto a heated substrate, where they condense and form
a thin film.
• Control and Monitoring: The procedure necessitates accurate control and monitoring
of the beam flux, substrate temperature, and additional parameters to attain the targeted
characteristics of the film.
• Layer-by-Layer Growth: MBE enables incremental layering in the growth of thin films,
allowing for meticulous regulation over the composition, thickness, and structure of the
deposited material.

3) Applications:

a) Semiconductor Devices: MBE is widely employed in the fabrication of semiconductor

devices, including quantum wells, heterostructures, and superlattices, because it can
precisely engineer material properties at the atomic scale.
b) Quantum Dots and Nanowires: It is applied in methods such as MBE for the
fabrication of quantum dots and nanowires, requiring precise management of material
properties crucial to their functionality.
430
 Challenges and Future Prospects of Thin Film Deposition Techniques: A Critical Review

c) Topological Insulators: MBE has contributed to the advancement of topological

insulators, a class of materials with unique electronic properties.

4) Advantages of MBE:

a) Precise Control: MBE provides exceptional control over the deposition process,
resulting in high-quality films with atomic precision.
b) Low Contamination: The high vacuum environment minimizes contamination during
film deposition, leading to pure, high-quality films.
c) Research and Development: MBE is extensively utilized in material research and
development for its capability to produce custom materials with specific properties.
Thermal evaporation is another commonly used technique for depositing thin films, and
it differs from Molecular Beam Epitaxy (MBE).

C. Thermal Evaporation:

Thermal evaporation entails heating a solid source material in a vacuum chamber until it
vaporizes. The vaporized material then condenses onto a substrate, creating a thin film. This
process usually occurs in a high-vacuum environment to avoid contamination and maintain
the purity of the deposited film.

1. Process of Thermal Evaporation:

a) Heating Source Material: The solid material (often a metal or semiconductor) is heated
to its evaporation temperature, causing it to transition directly from solid to vapor phase.
b) Deposition onto Substrate: The vaporized material moves toward a substrate, where it
condenses to create a thin film.
c) Controlled Thickness: The film thickness is regulated by the duration of the deposition
process and can be monitored in real-time using methods such as quartz crystal
microbalances or film thickness monitors.

2. Applications:

a) Thin Film Coatings: Thermal evaporation is employed to create coatings with a range
of properties, including optical coatings, protective layers, and conductive coatings.
b) Semiconductor Fabrication: It is used in the fabrication of thin film transistors, metal
contacts, and various semiconductor devices.

3. Advantages of Thermal Evaporation:

a) Cost-Effective: Thermal evaporation presents a comparatively cost-effective approach

for thin film deposition in contrast to methods such as MBE.
b) Uniform Coatings: It can produce uniform coatings over extensive areas, making it
ideal for industrial-scale applications.

431
International Journal of Research and Analysis in Science and Engineering

D. Arc Vapor Deposition (AVD):

1. Process of Arc Vapor Deposition:

a) Arc Generation: A high-voltage electrical discharge creates an arc between the target
material (cathode) and an anode, leading to the vaporization of the target material.
b) Film Deposition: The condensed vapor material forms a thin film as it settles onto a
substrate.
c) Controlled Fabrication: The procedure allows for exact regulation of parameters such
as arc current, voltage, and deposition time to attain the targeted characteristics of the
film.

2. Applications:

a) Wear-Resistant Coatings: Arc vapor deposition is used to produce wear-resistant

coatings on tools, machine components, and industrial equipment, enhancing their
longevity and performance.
b) Decorative Coatings: This method is utilized in creating decorative coatings for
architectural, automotive, and consumer goods, offering both aesthetic enhancement
and protection.
c) Tribological Coatings: This method is employed in the fabrication of tribological
coatings designed to minimize friction and wear on mechanical components and
surfaces.

3. Advantages of Arc Vapor Deposition:

a) High Adhesion and Density: Arc vapor deposition yields dense and adherent coatings
because the kinetic energy of the vaporized particles enhances their impact on the
substrate.
b) Uniformity: The process can produce highly uniform coatings over extensive substrate
areas, making it well-suited for industrial-scale applications.
c) Range of Materials: It can be utilized with a variety of target materials, including
metals, ceramics, and certain polymers, providing versatility in film composition.

E. Sputtering:

Sputtering, widely employed for thin film deposition, entails bombarding a target material
with high-energy particles to eject atoms and create a thin film on a substrate. The following
presents a comprehensive examination of the sputtering process, its diverse applications,
and its advantages.

Sputtering for Thin Film Deposition:

1. Process of Sputtering:

a) Bombardment of Target Material: High-energy ions or electrons are used to bombard

a target material, causing atoms to be ejected from the target surface.

432
 Challenges and Future Prospects of Thin Film Deposition Techniques: A Critical Review

b) Thin Film Formation: The atoms that are ejected travel towards the substrate where
they condense, resulting in the formation of a thin film that mirrors the composition of
the target material.
c) Control over Properties: The process of sputtering allows for meticulous control of
parameters such as gas pressure, target material, and substrate temperature, thereby
customizing the properties of the film.

2. Applications:

a) Optical Coatings: Sputtering is essential in the production of optical coatings on lenses,

mirrors, and other optical components in applications like cameras, telescopes, and
microscopes.
b) Solar Cells: The method is utilized in the fabrication of thin film photovoltaic cells,
which requires precise control over material properties and film uniformity is crucial
for efficiency.

3. Advantages of Sputtering:

a) Uniform Film Deposition: Sputtering has the capability to produce highly uniform and
conformal coatings, making it well-suited for applications that require precision in
controlling both thickness and composition.
b) Material Versatility: This technique can be utilized across a diverse range of materials,
encompassing metals, insulators, semiconductors, and transparent conductive oxides.
c) High-Density Films: The sputtering process can produce dense, high-quality films that
adhere well to the substrate.

F. Magnetron Sputtering:

This technique incorporates a magnetron to enhance the sputtering process, where atoms
are ejected from a target material by high-energy particle bombardment. It can be applied
with a wide range of materials.

1. Operation of Magnetron Sputtering:

a) Target Material: The material intended for deposition as a film is positioned as the
target within the vacuum chamber.
b) Creation of Plasma: In magnetron sputtering, a magnetic field is employed to trap
electrons, resulting in the creation of a high-density plasma. This boosts the sputtering
process by elevating the kinetic energy of the sputtered atoms.
c) Sputtering Process: Argon gas is introduced into the chamber, where the plasma
accelerates argon ions to bombard the target material, leading to the ejection of atoms
from the target.
d) Film Deposition: The atoms that have been ejected travel towards the substrate and
condense, resulting in the formation of a thin film. The substrate, positioned in front of
the target material, collects the sputtered atoms, allowing them to accumulate and form
the intended thin film.

433
International Journal of Research and Analysis in Science and Engineering

2. Advantages of Magnetron Sputtering:

a) Uniform Coating: Magnetron sputtering generally produces a more uniform film

thickness compared to other deposition techniques.
b) Control over Film Properties: The deposition parameters can be adjusted to control the
properties of the deposited film, such as thickness, composition, and microstructure.

3. Applications:

Magnetron sputtering finds extensive use in various industries, including:

• Semiconductor manufacturing
• Solar cell production
• Display and touch panel coating
• Architectural glass and automotive glass coating
• Data storage media production

G. Ion Beam Sputtering:

Ion Beam Sputtering is a physical vapor deposition (PVD) method that entails bombarding
a target material with a high-energy ion beam. Subsequently, the atoms or molecules ejected
from the target are deposited onto a substrate to generate a thin film.

1. Operation of Ion Beam Sputtering:

a) Ion Source: In Ion Beam Sputtering, an ion source generates a beam of ions, typically
inert gases like argon or oxygen.
b) Target Material: The target material to be sputtered is positioned inside the vacuum
chamber.
c) Ion Bombardment: The high-energy ions generated by the source are directed towards
the target material. The ions bombard the target surface, causing atom displacement and
ejection of material.
d) Deposition: The sputtered atoms from the target material travel to the substrate and
condense to create a thin film with the desired properties.

2. Advantages of Ion Beam Sputtering:

a) Controlled Energy: Ion Beam Sputtering enables precise control over the energy of the
ions used for bombardment, enabling tailored film properties.
b) High Purity Films: The process can produce high-purity films with low levels of
impurities, suitable for applications requiring superior cleanliness.

3. Applications of Ion Beam Sputtering:

a) Precision Optics: Used in fabrication of precision optical coatings for lenses and
mirrors.

434
 Challenges and Future Prospects of Thin Film Deposition Techniques: A Critical Review

b) Semiconductor Industry: Used in manufacturing of semiconductor devices and thin

film transistors.

H. Pulsed Laser Deposition (PLD):

Pulsed Laser Deposition is an exceptionally precise physical vapor deposition (PVD)

method that employs a high-energy pulsed laser to ablate a target material and subsequently
apply it as a thin film onto a substrate. Let's explore the details of Pulsed Laser Deposition.

1. Operation of Pulsed Laser Deposition:

a) Pulsed Laser: A high-energy laser is directed at the target material within a vacuum
chamber, leading to the rapid vaporization and ionization of the material.
b) Plume Formation: This process creates a plume of vaporized target material, consisting
of energetic atoms, ions, and clusters.
c) Film Deposition: The energetic species from the plume moves across the vacuum
chamber and condenses onto the substrate, creating a thin film.

2. Advantages of Pulsed Laser Deposition:

a) Stoichiometric Control: PLD enables accurate control of stoichiometry, making it ideal

for depositing complex compounds and multicomponent materials with the desired
composition.
b) Versatility: It can apply a broad variety of materials.

I. Direct current (DC) sputtering:

Let’s delve into the specifics of DC sputtering:

1. Operation of DC Sputtering:

a) Target Material: The material intended for deposition as a thin film is positioned as the
target inside a vacuum chamber.
b) Gas Introduction: In DC sputtering, an inert gas, like argon with low pressue is
introduced into the chamber.
c) Ionization of Gas: A direct current is applied to the target material, producing a
potential difference that leads to the ionization of the inert gas.
d) Sputtering Process: The positively charged ions of the inert gas bombard the target
material, causing atoms to be ejected from the target.
e) Deposition: The ejected atoms move through the vacuum chamber and deposit onto the
substrate, creating a thin film with the desired characteristics.

2. Advantages of DC Sputtering:

a) High Deposition Rate: DC sputtering can achieve relatively high deposition rates,
making it suitable for industrial-scale production.

435
International Journal of Research and Analysis in Science and Engineering

b) Simple Setup: The equipment required for DC sputtering is relatively simple, making
it cost-effective and easier to maintain.

3. Applications of DC Sputtering:

a) Semiconductor Industry: It is used to deposit thin films in integrated circuit

manufacturing.
b) Glass Coatings: It finds application in depositing transparent conducting oxide (TCO)
coatings on glass for applications in solar panels, display technologies, and architectural
glass.
c) Data Storage: DC sputtering is employed in the manufacturing of thin film magnetic
recording media for data storage.

J. RF Sputtering:

RF sputtering, also known as radio frequency sputtering, is a physical vapor deposition

(PVD) technique that employs radio frequency energy to sputter a target material and then
deposit it as a thin film onto a substrate. Renowned for its ability to provide exact control
over film properties, this method is widely embraced across various industries. Here, we
will outline the key characteristics of RF sputtering:

1. Operation of RF Sputtering:

a) Target Material: The material intended for deposition as a thin film is positioned as the
target within a vacuum chamber.
b) Gas Introduction: An inert gas, typically argon, is introduced into the chamber at a low
pressure.
c) Application of Radio Frequency: A radio frequency (RF) power source is employed to
ionize the inert gas, creating a plasma within the chamber.
d) Ion Bombardment: The positively charged ions in the plasma are driven towards the
target material by the applied electric field. These ions collide with the target, causing
atoms to be ejected from it.
e) Deposition: The ejected atoms move through the vacuum chamber and deposit onto the
substrate, forming a thin film with the desired characteristics.

2. Advantages of RF Sputtering:

a) Controlled Film Properties: RF sputtering provides precise control over film

composition, structure, and thickness, making it ideal for various applications.
b) High Uniformity: It can provide high uniformity in coating thickness across large
substrate areas, ensuring consistent film properties.
c) Low Substrate Heating: The minimal substrate heating during RF sputtering makes it
ideal for depositing thin films on heat-sensitive materials.

3. Applications of RF Sputtering:

a) Semiconductor Devices: RF sputtering is widely employed in production of integrated

circuits, diodes, and transistors.
436
 Challenges and Future Prospects of Thin Film Deposition Techniques: A Critical Review

b) Optical Coatings: It is used for depositing optical coatings on lenses, mirrors, and other
optical components.
c) Solar Cells: RF sputtering is employed in the manufacturing of thin film solar cells,
which play a vital role in renewable energy technologies.

K. Evaporation:

1. Mechanism: In evaporation, the source material is heated to produce vapor, which

subsequently condenses onto a substrate to create a thin film.
2. Energy Source: Heat energy is primarily used to vaporize the source material, either
by resistive heating, electron beam heating, or thermal evaporation.
3. Directionality: The vapor stream moves in straight lines from the source to the
substrate, leading to line-of-sight deposition, which can be less effective for coating
substrates with complex geometries.
4. Uniformity: Evaporation may exhibit less uniform film thickness distribution for
complex or large-area substrates due to its line-of-sight nature.
5. Applications: It is frequently used for depositing metals, metal oxides, and organic
materials in applications like optoelectronics, decorative coatings, and semiconductor
devices.

L. Sputtering:

1. Mechanism: Sputtering involves bombarding a target material with energetic particles,

typically ions, to eject and deposit atoms onto a substrate.
2. Energy Source: The bombardment energy is commonly delivered through various
methods, including direct current (DC), radio frequency (RF), magnetron, or ion beam
sputtering.
3. Directionality: Sputtering can offer more isotropic (non-line-of-sight) coverage,
enabling more effective coating of complex and three-dimensional substrates.
4. Uniformity: Because it can eject atoms from a target in multiple directions, sputtering
can achieve greater film thickness uniformity across different substrate geometries.
5. Applications: Sputtering is extensively used in the semiconductor industry, optical
coatings, precision optics, photovoltaics, and magnetic storage media due to its
capability to deposit a diverse range of materials while maintaining precise film
properties.

IV Conclusion:

In summary, although both evaporation and sputtering are physical vapor deposition (PVD)
techniques for applying thin films, they differ in their mechanisms, directionality, and
uniformity. Evaporation uses heat to vaporize the source material, often leading to line-of-
sight deposition, while sputtering ejects atoms from a target using energetic bombardment,
resulting in more isotropic coverage and better performance on complex substrates. Each
method has its advantages and is selected based on the specific needs of the thin film
deposition process.

437
International Journal of Research and Analysis in Science and Engineering

References:

1. Mandakini N. Chaudhari, Rajendrakumar B. Ahirrao, Sanabhau D. Bagul, “Thin film

Deposition Methods: A Critical Review,” International Journal for Research in Applied
Science & Engineering Technology (IJRASET) ISSN: 2321-9653; IC Value: 45.98; SJ
Impact Factor: 7.429 Volume 9 Issue VI June 2021.
2. D. A. Jameel, “Thin Film Deposition Processes”, International Journal of Modern
Physics and Applications Vol. 1, No. 4, pp. 193-199, 2015.
3. J. Thirumalai, Introductory Chapter: The Prominence of Thin Film Science in
Technological Scale, http://dx.doi.org/10.5772/67201.
4. A. Baptista, F. Silva, J. Porteiro, J. Míguez and G. Pinto, “Sputtering Physical Vapour
Deposition (PVD) Coatings: A Critical Review on Process Improvement and Market
Trend Demands, Coatings”, 8, 402; doi:10.3390/coatings8110402, 2018.
5. M. Urbina et.al; The theologies and strategies for the development of novel material
systems and coatings for applications in extreme environments: a critical review’
Manufacturing Rev. 5, 9 (2018).
6. Mattox, Donald M. "The Foundations of Vacuum Coating Technology" Noyes
Publications (2003).
7. Mattox, Donald M. and Vivivenne, Harwood Mattox (editors) "50 Years Of Vacuum
Coating Technology and the Growth of the Society of Vacuum Coaters", Society of
Vacuum Coaters (2007).
8. H. Soonmin, S. A. Vanalakar, Ahmed Galal and Vidya Nand Singh, A review of
nanostructured thin films for gas sensing and corrosion protection, Mediterranean
Journal of Chemistry, 7(6), 433-451, 2019.
9. P. A. Savale, Physical Vapor Deposition (PVD) Methods for Synthesis of Thin Films:
A Comparative Study, Archives of Applied Science Research, 8 (5):1-8,2016.

438
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

46. Designing the Arduino Based Nutrition Feeding

Hydroponic System
Niveditha N. Shahapur, Sapana S. Lalaki
Dept. EEE, KLS VDIT, Haliyal.
Allamprabhu V. Kolaki
Asst. Prof. KLS VDIT, Haliyal.

ABSTRACT:

Hydroponics, a modern farming technique that eliminates the need for soil, employs water
enriched with nutrients to facilitate plant growth. This study focuses on developing an
automated system for distributing water and nutrients to plants, equipped with sensors to
monitor water levels and nutrient concentrations. The system includes a TDS sensor for
measuring electrical conductivity, an Arduino UNO R3 for processing, and a proximity
sensor for detecting water levels.

KEYWORDS:

Hydroponics, TDS, Arduino UNO.

I. Introduction:

The Extreme weather changes and polluted soil and air can cause problems for plants grown
in open fields. To tackle this, hydroponics is becoming popular in farming. It doesn't use
soil; instead, it uses water mixed with nutrients for plants to grow. These nutrients are
important for plant growth and are measured using something called Electrical Conductivity
(EC). EC tells farmers how much nutrients are in the water for plants to absorb. If EC is too
low, plants won't grow well, but if it's too high, plants might not be able to absorb the
nutrients properly, causing problems like poisoning. Monitoring EC helps farmers make
sure plants get the right amount of nutrients to grow well

For plants to grow well, they need the right water, nutrients, and enough air. These factors
help plants absorb nutrients, which they use to make energy. Roots use this energy to take
in water and more nutrients from the water they're in. This whole process helps plants grow
better. But if there's too much water, there might not be enough oxygen for the roots, which
can slow down growth.

439
International Journal of Research and Analysis in Science and Engineering

Using technology, like computers, helps make things easier. Automated systems can help
monitor and control these factors, making sure plants get what they need to grow
healthy and strong.

The development of science and technology, especially computer technology occurs

rapidly. The scope of application of technology is also increasingly widespread, covering
various aspects of human life so that efforts to meet human needs are increasing and
complex. Computer technology can be positively correlated to improving the quality of life
and human well-being. Rapid and accurate system automation is an important requirement
that encourages the creation of automated tools. Devices that are integrated with various
peripherals and other supporting devices make the performance of these technologies
become more applicable and functional

In 2014, Qalyubi, Pudjojono and Widodo [8] conducted an experiment on the effects of
water discharge and the provision of nutrients to the growth of kale plants (Ipomoea
aquatica forsk) in the hydroponics NFT system. The results revealed that the EC or the
concentration of hydroponic nutrition is very influential on the growth of kale plants. EC
amounted to 460 ppm indicates the deficit of solution concentration, so that the kale plants
grow slowly.

II. Method:

This section talks about how to make a system, like the one for hydroponics. There are four
main steps: figuring out what the system needs to do, designing it, building it, and testing
it.

First, you gather all the information about what the system should do, like picking the right
sensors. This step also includes looking at experiments and research. The outcome of this
step is a list of what the system should do, both the main things (functional) and other details
(non-functional).

The main things the system should do include keeping the right level of nutrients in the
water, adding water when needed, and reading data from sensors. It should also
automatically add the right fertilizers and check the levels of nutrients in the water. Plus, it
should adjust things like how much fertilizer to add.

The other details include what parts to use, like a specific microcontroller and power supply.
The system should also work without needing a person to control it, and it should connect
directly to the hydroponic system.

The second step is designing the system. This means taking all the information from the
first step and figuring out how to put it together, both in hardware (like the physical parts)
and software (like the computer programs). The result of this step is a plan showing how
everything will work together.

In simple terms, during the third stage of system development, called system
implementation, the actual hardware and software are put together and made ready to work.
In this specific project, they're using an Arduino Uno R3 board, which is like the brain of
440
 Designing the Arduino Based Nutrition Feeding Hydroponic System

the system. It has a microcontroller, which is like a tiny computer, along with some special
sensors like the SHARP GP2Y0A21 and TDS sensor, and a servo motor actuator. These
sensors help to detect things like the water level and the quality of the hydroponic nutrient
solution in a container. Based on this information, the system decides when to turn on or off
the water or fertilizer faucets using the servo motor.

To put it simply, "close the faucet head" means to shut off the water or fertilizer flow. In
this project, they're using a software called Arduino IDE, which is like a tool to tell the
system what to do. It's installed on a computer running Windows 10, and it helps to program
the hardware, like telling the system when to open or close the faucets based on the
information from the sensors.

The last stage of system development is called system testing. Here, they collect data to
check if both the hardware and software are working correctly. They look at the information
gathered from earlier stages, like the design and implementation phases, and see if
everything matches up. They run tests to make sure input devices, like sensors, and external
devices, like the servo motor, are working properly. They also check if the system does what
it's supposed to do based on its requirements, like turning on or off the faucets at
the right times.

III. Results and Analysis:

After going through all the steps of building the system, the next thing to do is to gather all
the data from the tests and describe what happened. This includes discussing and analysing
the results. Beforehand, the system's design was made, which includes diagrams showing
how the hardware parts are connected and how the software works.

Figure 1: Block diagram of hydro phonics

The software part also has diagrams, like flowcharts, to show how different parts of the
program work. For instance, Figure 2 shows a flowchart for the setup part of the software,
which is called "void setup ()". It's like setting up the stage before the play begins.

441
International Journal of Research and Analysis in Science and Engineering

Figure 2: Flowchart of system initialisation

Next, the void loop () function, which is a function that will run continuously, contains a
call to drive the main function of the system. Figure 3 shows the flow chart of the function
of the void loop () system.

Figure 3: Flow Chart of Hydro Phonics System

The hardware implementation means putting together all the physical parts that have been
made. So, in this case, they've used an Arduino Uno R3 board, which is like the brain of the
system. It's connected to different sensors: one to measure how much stuff is in the water
(TDS sensor), another to see how high the water is (distance sensor), and three actuators
that move things around (servo actuators). So, basically, they've built a system using these
parts to control and monitor things.

442
 Designing the Arduino Based Nutrition Feeding Hydroponic System

Set up the hardware directly onto a small-scale model of a hydroponic system with 12 holes
for plants. To connect everything, they've used a prototype shield that plugs into the Arduino
Uno R3 board. This shield has enough pins to attach all the parts they need for the system.
They've put the Arduino board and the prototype shield on top of the container that holds
the nutrients for the plants. So, everything is connected and mounted in one place, like in
the picture they've shown in Figure.

Figure A: Hydroponics System Setup

Figure B: Circuit Diagram of Hydro Phonics

443
International Journal of Research and Analysis in Science and Engineering

Figure C: Seedling Setup

Implementing software like building a robot using a special toolkit called Arduino IDE
version 1.6.5. You use a language called C to tell the robot what to do. The software is made
by writing instructions for the robot in three main parts:

1. Initial declaration This part sets up everything the Arduino needs, like the tools it will
use.
2. Void setup () function Here, you tell the Arduino what to do when it starts up, like
turning on its sensors.
3. void loop () function This is where you tell the Arduino what to do over and over again,
like moving around or checking for obstacles.

So, when you implement software, you're basically giving instructions to a system, step by
step, to make it do what you want.

Initializing and declaring global variables Before we start, we need to set things up and tell
our system about some important stuff that we'll be using throughout the process.

Void setup () function This is like the setup phase where we get everything ready. In this
case, it's about connecting our device (like the Arduino) and making sure it can talk to the
computer. This part also includes:

• Declaring which pins on the Arduino will be connected to different things.

• Turning on communication with sensors and motors, like TDS sensors, GP2Y0A21
sensors, and a servo motor.

Testing the system First, we check each sensor and input devices to make sure they're
working correctly. Then, we test the output devices to see if they respond as expected.
Finally, we do an overall test to make sure the whole system does what it's supposed to do.

Testing the TDS sensor to check the TDS sensor, we compare its readings with a known
device called a TDS meter. We convert the sensor's analog voltage into parts per million
(PPM) using a formula.
444
 Designing the Arduino Based Nutrition Feeding Hydroponic System

This formula uses the sensor's voltage output and compares it with the value measured by
the TDS meter. We repeat this test 10 times, placing the sensor and the TDS meter close
together. Then, we plot the data on a graph to see how well the sensor matches the
TDS meter readings

Testing the outlet device (servo motor) We check if the output of the servo motor matches
what we told it to do. In this system, we only have one outlet device, which is a servo motor.
This motor controls the faucet of containers for water and fertilizer.

Servo motor testing We want to make sure the servo motor works correctly. We test it by
commanding it to move to two positions:

• Open the faucet (move to position 30 degrees).

• Close the faucet (move to position 100 degrees).

Testing the whole system: After checking the inlet (like sensors) and outlet (like motors)
devices separately, we test the entire system altogether. The whole system is made up of
parts that we've already tested individually. Once we put all these parts together, we can test
how well they work as a complete system.

Comparative Analysis:

Table 1: Hydroponic Averages Compared with Ordinary Soil Yields

Name of Crop Hydroponic Equivalent per Acre Agricultural Average per Acre
Wheat 5,000 lb. 600 lb.
Oats 3,000 lb. 850 lb.
Rice 12,000 lb. 750-900 lb.
Maize 8,000 lb. 1,500 lb.
Soybean 1,500 lb. 600 lb.
Potato 70 tons 8 tons lb.
Beet root 20,000 lb. 9,000 lb.
Cabbage 18,000 lb. 13,000 lb.
Peas 14,000 lb. 2,000 lb.
Tomato 180 Tones 5-10 Tones
Cauliflower 30,000 lb. 10-15,000 lb.
French bean 42,000 lb. of pods for eating -
Lettuce 21,000 lb. 9,000 lb.
Lady’s finger 19,000 lb. 5-8,000 lb.
Cucumber 28,000 lb. 7,000 lb.

445
International Journal of Research and Analysis in Science and Engineering

IV Conclusion:

This study aims to explore how to make a small-scale automatic NFT hydroponic system
using TDS sensors. The research found that this system can accurately measure the level of
solutes in the nutrient solution using TDS sensors, which are measured in parts per million
(PPM). The system can automatically add nutrients to maintain a minimum concentration
of 800 ppm in the solution. The TDS sensor, when converted from voltage units to PPM,
showed an accuracy of 97.8%. The researchers also developed software to control the
hardware components based on the system's requirements, Overall, the designed system
meets all the planned functional requirements for automated nutrient delivery in
NFT hydroponic

References:

1. A. D. Susila, “System Hydroponic,” in Dasar-dasar Horticulture, 2013, pp. 1–21.

2. B. Perwtasari, M. Tripatmasari, and C. Wasonowati, “Pengaruh Media Tanam dan
Nutrisi terhadap Pertumbuhan dan Hasil Tanaman Pakchoi (Brassica juncea L.) dengan
Sistem Hidroponik,” J. Trunojoyo, vol. 5, no. 1, pp. 14–25, 2012.
3. Gusmardi, “Aplikasi Pupuk A-B Mix dengan Konsentrasi Tertentu pada
Tanaman Salada Hijau (Lactuca sativa) secara Hidroponik di PT. Parung Farm Bogor,”
Universitas Andalas, 2013.
4. I. S. Roidah, “Pemanfaatan Lahan Dengan Menggunakan Sistem Hidroponik,” J. Univ.
Tulungagung, vol. 1, no. 2, pp. 43–50, 2014.
5. S. Asyiah, “Kajian Penggunaan Macam Air dan Nutrisi pada Hidroponik Sistem DFT
terhadap Pertumbuhan dan Hasil Baby Kailan,” Universitas Sebelas Maret, 2013.
6. M. Kautsar, Rizal Isnanto, and E. D. Widianto, “System Monitoring Digital
Penggunaan dan Kualitas Kekeruhan Air PDAM Berbasis Mikrokontroler ATMega328
Menggunakan Sensor Aliran Air dan Sensor Fotodiode,” J. Teknol. Dan Sist. Komput.,
vol. 3, no. 1, 2015.
7. M. Diansari, “Pengaturan Suhu, Kelembaban, Waktu Pemberian Nutrisi dan Waktu
Pembuangan Air untuk Pola Cocok Tanam Hidroponik berbasis Mikrokontroler AVR
ATMega 8535,” Universitas Indonesia, 2008.
8. I. Qalyubi, M. Pudjojono, and S. Widodo, “Pengaruh Debit Air dan Pemberian Jenis
Nutrisi terhadap Pertumbuhan Tanaman Kangkung pada Sistem Irigasi Hidroponik
NFT (Nutrient Film Technique),” J. Teknol. Pertan. Univ. Jember, vol. 1, pp. 2–6, 2014.
9. P. Lingga, Hidroponik, Bercocok Tanam Tanpa Tanah.Jakarta: Penebar Swadaya,
2005. [10] “Arduino Uno and Genuino Uno,” 2016. [Online]. Available:
https://www.arduino.cc/en/Main/ArduinoBoard Uno. [Accessed: 12-Nov-2016].
10. Anonimous, “SHARP GP2Y0A21YK0F,” 2006.
11. Marliana and A. Wahjudi, “Rancang Bangun Perangkat Lunak Unit Kontrol Alat Ukur
Sudu Cross Flow Water Turbine berbasis Pengolahan Citra,” J. ITS, vol. 3, no. 2,

446
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

47. Multilingual Regional Speech Classification

Using Recurrent Neural Networks
Sunilkumar M. Hattaraki, Ranjeeta Kumbar,
Sushma M. Aloor, Ranjeeta Patil, Ramya Vajjaramatti
Department of E&CE
B.L.D.E.A’s V.P.Dr. P.G.Halakatti.College of Engineering and Technology,
Vijayapura, India, Visvesveraya Technological University,
Belagavi, Karnataka, India.
Shankarayya G. Kambalimath
Department of E&CE, Basaveshwar Engineering College,
Bagalkote, Karnataka, India. Visvesveraya Technological University,
Belagavi, Karnataka, India.

ABSTRACT:

This study addresses the classification of ten regional languages using Recurrent Neural
Networks (RNNs) based on Mel-Frequency Cepstral Coefficients (MFCCs) extracted from
audio recordings. The languages under consideration include Kannada, Hindi, Marathi,
Telugu, Malayalam, Bengali, Gujarati, Punjabi, Tamil, and Urdu. Motivated by the need
for robust multilingual speech recognition systems that can accommodate linguistic
diversity, this research aims to explore the effectiveness of RNNs in handling diverse
language datasets. Despite advances in speech recognition, there remains a research gap
in the development of models that accurately classify a wide range of regional languages
using limited training data. This study addresses this gap by leveraging deep learning
techniques and extensive data preprocessing to enhance classification accuracy. The RNN
architecture comprises two LSTM layers augmented with batch normalization and dropout
layers for regularization. Experimental results demonstrate promising outcomes with a
training accuracy of 73.00% and a test accuracy of 64.33%, showcasing the model's
capability to distinguish between the diverse phonetic and linguistic features of the selected
languages. Applications of this research include enhanced speech recognition systems for
diverse linguistic communities, automated language identification in multilingual
environments, and preservation of linguistic heritage through technology-driven
approaches. This study underscores the potential of RNNs in advancing the field of
multilingual speech processing and contributes valuable insights into addressing the
challenges of language diversity in artificial intelligence applications.

447
International Journal of Research and Analysis in Science and Engineering

KEYWORDS:

Regional Languages, Recurrent Neural Networks (RNNs), Mel-Frequency Cepstral

Coefficients (MFCCs), Speech Recognition and Multilingual Classification.

I. Introduction:

The rapid advancement of speech recognition technology has significantly transformed the
way we interact with machines and digital systems. From voice-activated assistants to
automated customer service systems, speech recognition has become an integral part of our
daily lives. However, the majority of these systems are predominantly tailored to handle a
limited number of globally dominant languages, such as English, Spanish, and Mandarin.
This limitation poses a significant challenge in linguistically diverse regions where a
multitude of languages and dialects are spoken.

India, for instance, is a country with a rich tapestry of languages, each with its unique
phonetic and linguistic characteristics. The inability of current speech recognition systems
to effectively process and classify regional languages limits their applicability and utility in
such multilingual environments. There is a pressing need for robust multilingual speech
recognition systems that can accurately recognize and classify a wide range of regional
languages. This study aims to address this gap by exploring the potential of Recurrent
Neural Networks (RNNs) in the classification of ten regional languages using audio
recordings.

In the contemporary world, security remains a paramount concern across various sectors,
particularly in financial institutions such as banks. The rise in theft incidents and
sophisticated intrusion techniques necessitates the development of advanced security
systems that can effectively safeguard valuable assets. Traditional security measures,
although essential, often fall short in providing real-time detection and prevention
capabilities. Consequently, there is a compelling need for innovative security solutions.

The primary objective of this research is to develop and evaluate a robust RNN-based model
for the classification of ten regional languages, namely Urdu, Kannada, Hindi, Marathi,
Telugu, Malayalam, Bengali, Gujarati, Punjabi, and Tamil. The study leverages Mel-
Frequency Cepstral Coefficients (MFCCs) as the primary feature extraction technique from
audio recordings. The specific objectives of this research are:

To design and implement an RNN architecture that can effectively classify audio recordings
of ten regional languages.

To preprocess and augment the audio data to enhance the performance of the RNN model.

To evaluate the model's performance in terms of training and test accuracy.

To identify potential applications of the developed system in multilingual speech

recognition and other related domains.

448
 Multilingual Regional Speech Classification Using Recurrent Neural Networks

Despite the significant strides made in the field of speech recognition, there remains a
noticeable gap in the development of models that can accurately classify a wide range of
regional languages using limited training data. Most existing models are either not trained
on diverse language datasets or lack the necessary robustness to handle linguistic variations
effectively. This research seeks to fill this gap by employing deep learning techniques,
specifically RNNs, which are well-suited for sequence modeling tasks such as speech
recognition. The study also emphasizes extensive data preprocessing and augmentation to
overcome the challenges posed by limited and diverse training data.

The contributions of this research are manifold. Firstly, it provides a comprehensive

evaluation of the effectiveness of RNNs in the classification of regional languages, offering
valuable insights into the potential and limitations of such models. Secondly, it highlights
the importance of data preprocessing and augmentation in enhancing model performance,
which can serve as a reference for future studies in this domain.

The applications of this research are far-reaching. Enhanced speech recognition systems
that can cater to diverse linguistic communities will significantly improve accessibility and
user experience in multilingual environments. Automated language identification systems
can be employed in various sectors, including customer service, education, and healthcare,
to provide tailored services based on the detected language. Furthermore, the preservation
of linguistic heritage through technology-driven approaches will ensure that regional
languages continue to thrive in the digital age.

II. Literarure Survey:

Burchi et al. [1] introduced a model that sets a new state-of-the-art performance on the LRS3
dataset, achieving a Word Error Rate (WER) of 0.8%. Additionally, on the newly
established MuAViC benchmark, their model demonstrates an absolute average-WER
reduction of 11.9% compared to the original baseline. Notably, this model is capable of
performing audio-only, visual-only, and audio-visual speech recognition during testing,
showcasing its versatility and effectiveness.

Gupta et al. [2] developed a CNN-based Automatic Speech Recognition System (ASRS)
that models raw speech signals. The speech corpus, created by the researchers, includes
recordings in Hindi, English, Punjabi, and Bengali, performed by 50 male native speakers
of Hindi and Punjabi who could also speak English and Bengali. Using Mel-Frequency
Cepstral Coefficient (MFCC) for feature extraction, they designed a 2D CNN model with
six layers to recognize speech samples in each language.

Praveen et al. [3] describe the SRI-B systems proposed for task-1 of the inaugural MERLIon
CCS challenge in both closed and open domains. For the closed task, they used an end-to-
end conformer architecture trained for automatic speech recognition (ASR) with RNN-T
loss, later adapted for language classification. This system achieved a 13.9% Equal Error
Rate (EER) and 81.7% Balanced Accuracy (BAC) on the evaluation set. For the open track,
an ensemble of OpenAI’s Whisper model and one of the ASR models from the closed track
was used, achieving 9.5% EER and 78.9% BAC. Compared to the challenge baseline, the
closed track system showed a 35.9% relative improvement in EER, and the open track
system showed a 56.2% improvement.
449
International Journal of Research and Analysis in Science and Engineering

Parashar et al. [4] present a system for multilingual speech sentiment recognition using
spiking neural networks (SNNs), trained with Mel-frequency cepstral coefficients (MFCC)
features. Experiments were conducted on a combined dataset of SAVEE (English), EMO-
DB (German), EMOVO (Italian), and CaFE (French). The results show that SNNs achieved
76.85% accuracy, outperforming traditional artificial neural networks (ANNs) like MLP
and RNN on this task.

Nuthakki et al. [5] proposed a method to address the complexities of voice synthesis with
minimal aperiodic distortion, making it suitable for communication recognition models.
Despite some audible flaws, the model closely mimics human speech. They emphasize the
need for incorporating sentiment analysis into text categorization using natural language
processing, considering the varying intensity of emotions across countries. To enhance
voice synthesis, more hidden layers and nodes should be added to the mixture density
network. The proposed algorithm requires a robust network foundation and optimization
methods for optimal performance. The paper aims to provide both experienced researchers
and beginners with insights into developing a deep learning approach, highlighting progress
in overcoming fitting issues with limited training data and the need for more input parameter
space in DL-based methods.

Abdal et al. [6] utilize a BiLSTM architecture to effectively capture both contextual and
sequential dependencies in text data. Their model combines word embeddings with
character-level embeddings to encapsulate semantic and morphological information in
comments. The model's performance is compared with several advanced methods, including
RNN and LSTM. Experimental results indicate that the proposed model excels in
classifying multilingual toxic comments, achieving a superior accuracy of 94.21%.

Athish et al. [7] aim to enhance transcription accuracy by addressing challenges like speaker
accents, background noise, and audio quality. They will test the system with diverse audio
sources to ensure reliable transcription of various speech types. A report will detail the
system's performance and limitations, accompanied by a working prototype. The findings
will benefit industries relying on voice recognition and transcription technologies,
improving communication, accessibility, and productivity.

Malik et al. [8] developed a framework comprising data preprocessing, RoBERTa-based

data representation, fine-tuning, and hope speech classification into two labels. They created
a new Russian corpus for hope speech detection from YouTube comments and conducted
experiments using semi-supervised bilingual English and Russian datasets. The framework
achieved benchmark performance, surpassing baseline methods. Specifically, the
translation-based approach using Russian-RoBERTa achieved the highest accuracy of 94%
and an F1-score of 80.24% in their experiments across English and Russian languages.

Mussakhojayeva et al. [9] developed a fine-tuning strategy for the Whisper model to
enhance performance specifically for the Turkic language family and maintain high
accuracy across high-resource languages. Their Soyle model demonstrated superior
performance across 11 Turkic languages and official United Nations languages. They also
introduced the first large open-source speech corpus for the Tatar language, TatSC, which
significantly improved Tatar speech recognition accuracy. Emphasizing noise robustness,
they have open-sourced both the model and TatSC to foster further research.

450
 Multilingual Regional Speech Classification Using Recurrent Neural Networks

Their approach sets a precedent for creating multilingual speech recognition models tailored
to low-resource language families.

Gutha et al. [10] outline the Code Fellas team's strategies for HASOC 2023 Task 4:
Annihilate Hate, focusing on extending hate speech detection to Bengali, Bodo, and
Assamese languages. Their methods leverage Long Short Term Memory (LSTM) with
Convolutional Neural Networks (CNN), alongside pre-trained Bidirectional Encoder
Representations from Transformers (BERT) models such as IndicBERT and MuRIL. Their
findings highlight the effectiveness of these approaches, achieving notable results:
IndicBERT attains a significant F1 score of 69.726% for Assamese, MuRIL achieves
71.955% for Bengali, and a BiLSTM model with an added Dense Layer achieves an
impressive 83.513% for Bodo.

III. Methodology:

The audio data set [11] with ten Indian languages comprises of audio recordings in Kannada,
Hindi, Marathi, Telugu, Malayalam, Bengali, Gujarati, Punjabi, Tamil, and Urdu is used for
training the model. Among the total 1300 audio files of ten regional languages, 1000 audio
files are used for training the model and 300 files are used for testing. This dataset is
designed for tasks related to speech recognition and language processing. Each language
category includes various audio samples recorded under different conditions, providing a
diverse set of speech patterns and accents. The dataset is structured to facilitate research and
development in multilingual speech analysis, with potential applications in automatic
speech recognition, sentiment analysis, and dialect identification. Its availability on Kaggle
ensures accessibility and encourages collaborative efforts in advancing speech technology
across diverse linguistic contexts.

A. Input: The process begins with a collection of audio files encompassing diverse
languages such as Kannada, Hindi, Marathi, Telugu, Malayalam, Bengali, Gujarati, Punjabi,
Tamil, and Urdu. Each audio file serves as the raw input for subsequent processing stages
aimed at classifying these languages effectively.

B. Data Preprocessing: Upon receiving the audio files, the first step involves loading each
file into the system. This ensures that the audio data is accessible for further manipulation
and feature extraction. Additionally, data augmentation techniques may be applied to
enhance the robustness and variability of the dataset. Techniques such as pitch shifting or
adding noise simulate different acoustic environments, thereby improving the model's
ability to generalize across varied conditions.

C. Feature Extraction (MFCC): The heart of the preprocessing phase lies in extracting
Mel-Frequency Cepstral Coefficients (MFCCs) from the audio signals. MFCCs are
computed to capture the spectral characteristics of each audio segment. This involves
breaking down the audio signal into short frames and computing the spectrum of each frame.
The resulting coefficients represent the power spectrum of the audio signal, quantifying its
frequency components. The Figure 1 illustrates the sequential steps of the proposed method
for multilingual regional speech classification using Recurrent Neural Networks (RNNs). It
starts with input audio files, undergoes data preprocessing including loading, augmentation,
and feature extraction using Mel-frequency Cepstral Coefficients (MFCC), resulting in a
451
International Journal of Research and Analysis in Science and Engineering

structured MFCC feature matrix. The RNN model architecture, comprising LSTM layers
with batch normalization and dropout for regularization, is trained using the feature matrix.
Model evaluation is performed with performance metrics, and classification results in
predicted language labels are obtained as output. This flowchart provides an overview of
the methodology employed to classify audio samples into their respective regional
languages.

D. Feature Extraction: The extracted MFCC features are structured into a matrix format.
Each audio sample is represented as a sequence of MFCC vectors, where each vector
contains coefficients representing different frequency bands. This structured representation
forms the input data that will be fed into the Recurrent Neural Network (RNN) for language
classification.

E. RNN Model Architecture: The RNN model architecture is designed to effectively

process the structured MFCC features. It begins with an Input Layer that accepts the
sequence of MFCC vectors. This is followed by two LSTM (Long Short-Term Memory)
layers, chosen for their ability to model sequential dependencies in temporal data like
speech. Batch Normalization layers stabilize and accelerate training by normalizing the
activations of the LSTM outputs. Dropout layers are introduced to prevent overfitting by
randomly dropping neurons during training. A Dense layer performs the final classification
based on the learned features, with another Dropout layer for regularization, and an Output
Layer produces predictions for the language labels based on the extracted features.

Figure 1: Flowchart of RNN Model

452
 Multilingual Regional Speech Classification Using Recurrent Neural Networks

This architecture is designed to effectively capture temporal dependencies in audio data

(MFCC sequences) and classify them into one of the ten regional languages considered in
your dataset. Adjustments in dropout rates, LSTM units, or dense layer sizes can further
optimize performance based on specific requirements and constraints of the dataset.

F. Model Training: Once the model architecture is established, it is trained using the
structured MFCC features and corresponding language labels. During training, the model
adjusts its parameters to minimize a specified loss function (such as categorical cross-
entropy) using an optimizer (e.g., Adam optimizer). This iterative process fine-tunes the
model's weights and biases to optimize its ability to classify the language of audio samples
accurately.

G. Model Evaluation: After training, the model's performance is evaluated using a separate
validation or test set of audio samples. Performance metrics such as accuracy, precision,
recall, and F1-score are computed to assess how well the model classifies the language of
the audio samples. These metrics provide insights into the model's effectiveness and its
ability to generalize to unseen data.

H. Output: Finally, the model produces classification results where each audio file is
assigned a predicted language label based on its MFCC features and the trained RNN model.
These predicted labels provide actionable insights into the linguistic content of the audio
files, demonstrating the efficacy of the RNN-based approach in multilingual speech
classification tasks.

Table 1: Performance of RNN Model on Regional Language Classification

Sl. No. RNN Model Accuracy

1. Training 73.00 %
2. Testing 64.33 %

Figure 2: Training Accuracy vs. Validation Accuracy

453
International Journal of Research and Analysis in Science and Engineering

This detailed explanation outlines each step of the flow chart diagram as shown in Figure
1, from initial data preprocessing and feature extraction to model architecture, training,
evaluation, and final classification output. Each component plays a crucial role in achieving
accurate and robust language classification using Recurrent Neural Networks (RNNs) and
Mel-Frequency Cepstral Coefficients (MFCCs) extracted from audio recordings.

IV. Results and Discussion:

The table 1 summarizes the performance of the RNN model in terms of accuracy for both
the training and testing datasets. The training accuracy is 73.00%, indicating that the model
performs well on the data it was trained on. The testing accuracy is slightly lower at 64.33%,
reflecting the model's performance on unseen data.

The Figure 2, shows the training and validation accuracy of the model over 20 epochs. The
training accuracy (blue line) steadily increases from around 0.2 to 0.7, indicating effective
learning.

Figure 3: Confusion Matrix of Implemented Model

The validation accuracy (orange line) also improves but with more fluctuations, reaching
about 0.6 by the 20th epoch. The gap between the two curves suggests some overfitting,
where the model performs better on the training data than on the validation data.

The Figure 3, presents a normalized confusion matrix that visualizes the performance of the
model in classifying 10 different languages. Each row represents the actual language, while
each column represents the predicted language. The diagonal elements show the percentage
of correct predictions for each language, with values close to 1.0 indicating high accuracy.
For example, Tamil (10) has the highest accuracy at 0.93, followed by Kannada (1) at 0.93.
The off-diagonal elements indicate misclassifications, where some languages like Bengali
(6) and Telugu (4) show notable confusion with others. This matrix helps in understanding
the strengths and weaknesses of the model in distinguishing between different languages.

454
 Multilingual Regional Speech Classification Using Recurrent Neural Networks

V. Conclusion and Future Scope:

This study focuses on classifying ten regional languages using Recurrent Neural Networks
(RNNs) based on Mel-Frequency Cepstral Coefficients (MFCCs) from audio recordings,
including Kannada, Hindi, Marathi, Telugu, Malayalam, Bengali, Gujarati, Punjabi, Tamil,
and Urdu. Despite advances in speech recognition, there remains a gap in models that
accurately classify a wide range of regional languages with limited training data. This
research leverages deep learning and data preprocessing to enhance classification accuracy,
achieving a training accuracy of 73.00% and a test accuracy of 64.33%. The proposed RNN
model, with two LSTM layers, batch normalization, and dropout layers, effectively
distinguishes the diverse phonetic and linguistic features of the selected languages.
Applications include improved speech recognition systems, automated language
identification, and the preservation of linguistic heritage.

Future work should expand the dataset to include more languages and dialects, integrate
advanced data augmentation techniques, and explore alternative architectures like
transformers. Addressing limited training data with transfer learning and semi-supervised
learning, and implementing real-time speech recognition systems could significantly impact
education, customer service, and accessibility tools. Collaborations with linguists and
cultural experts will ensure technological advancements support linguistic and cultural
preservation.

References:

1. M. Burchi, et al., "Multilingual Audio-Visual Speech Recognition with Hybrid

CTC/RNN-T Fast Conformer," in ICASSP 2024-2024 IEEE International Conference
on Acoustics, Speech and Signal Processing (ICASSP), 2024.
2. A. Gupta, et al., "An Approach to Recognize Speech Using Convolutional Neural
Network for the Multilingual Language," in 2023 Global Conference on Information
Technologies and Communications (GCITC), 2023.
3. K. Praveen, et al., "Language identification networks for multilingual everyday
recordings," 2023.
4. S. Parashar et al., "Multilingual Speech Sentiment Recognition Using Spiking Neural
Networks," in International Conference on Big Data Analytics, Cham: Springer Nature
Switzerland, 2023.
5. P. Nuthakki, et al., "Deep learning based multilingual speech synthesis using multi
feature fusion methods," ACM Transactions on Asian and Low-Resource Language
Information Processing, 2023.
6. M. N. Abdal, et al., "Multilingual Toxic Comment Classification Using Bidirectional
LSTM," in International Conference on Electrical and Electronics Engineering,
Singapore: Springer Nature Singapore, 2023.
7. A. Y. Athish, et al., "Multilingual Speech Recognition Using Reinforcement Learning,"
in 2023 14th International Conference on Computing Communication and Networking
Technologies (ICCCNT), 2023.
8. M. S. I. Malik, et al., "Multilingual hope speech detection: A Robust framework using
transfer learning of fine-tuning RoBERTa model," Journal of King Saud University-
Computer and Information Sciences, vol. 35, no. 8, pp. 101736, 2023.

455
International Journal of Research and Analysis in Science and Engineering

9. S. Mussakhojayeva, et al., "Noise-Robust Multilingual Speech Recognition and the

Tatar Speech Corpus," in 2024 International Conference on Artificial Intelligence in
Information and Communication (ICAIIC), 2024.
10. A. R. Gutha, et al., "Multilingual Hate Speech and Offensive Language Detection of
Low Resource Languages," in FIRE (Working Notes), 2023.
11. C. B. H., "Audio dataset with 10 Indian languages," Kaggle, 2021. [Online].
Available:https://www.kaggle.com/datasets/hbchaitanyabharadwaj/audio-dataset-
with-10-indian-languages. [Accessed: 25-Jul-2024].

456
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

48. Contactless Care Systems: A Review

Basavaraj Soratur
Assistant Professor,
Dept of Electronics and Communication Engineering.
SKSVMACET, Lakshmeshwar, Karnataka, India.
Mahabaleshwar S. Kakkasageri
Professor, Dept of Electronics and Communication Engineering.
Basaveshwar Engineering College Bagalkot, Karnataka, India.
Subhas Meti
Professor, Dept of Electronics and Communication Engineering.
SKSVMACET, Lakshmeshwar, Karnataka, India.

ABSTRACT:

Contactless care systems have gained significant attention in the healthcare industry due to
their ability to monitor vital signs and detect abnormalities without physical contact. This
paper aims to provide a thorough examination of the current landscape of contactless care
systems, with a particular emphasis on three key areas: fall detection, breathing rate
monitoring, and heart rate monitoring. The review delves into the various smart sensors
and wireless technologies employed in these systems, such as cameras, radar, and motion
detectors, and explores how they facilitate the continuous tracking of an individual's
movement, breathing patterns, and heart rate without the need for wearable devices.

The aging population and the increasing cost of healthcare have driven the need for
advanced monitoring systems that can provide continuous, real-time data on an individual's
health status. Wearable devices have emerged as a promising solution, as they can interact
with the human body and monitor various physiological parameters. However, the adoption
of wearable devices has been hindered by issues such as comfort, user compliance, and
potential data privacy concerns [1][2].

I Introduction:

The aging population and rising healthcare costs have driven the demand for advanced
monitoring systems that can provide continuous, real-time data on an individual's health
status [3]. Wearable devices have emerged as a promising solution, as they can interact with
the human body and monitor various physiological parameters [2][1].

457
International Journal of Research and Analysis in Science and Engineering

However, the adoption of wearable devices has been hindered by issues such as comfort,
user compliance, and potential data privacy concerns [4].

In contrast, contactless care systems have been developed to address these limitations by
leveraging a suite of sophisticated sensors that can capture vital signs and detect anomalies
without any physical contact with the individual. These systems utilize a variety of sensors,
including cameras, radar, and motion detectors, to track an individual's movement,
breathing patterns, and heart rate [1][3].

One of the key applications of contactless care systems is fall detection. Falls are a
significant concern for the elderly population, particularly those living independently, as
they can result in serious injuries, hospitalization, and even mortality [5]. By continuously
monitoring an individual's movement and quickly identifying potential falls, contactless fall
detection systems can enable a rapid response and increased safety for vulnerable
populations, such as the elderly living independently [5].

In addition to fall detection, contactless care systems also play a crucial role in the
monitoring of vital signs, such as breathing rate and heart rate. These systems can provide
valuable insights into an individual's overall health status and help healthcare professionals
detect potential issues before they escalate into more serious conditions.

This paper presents a comprehensive review of the current state of contactless care systems,
with a focus on fall detection, breathing rate monitoring, and heart rate monitoring. The
review covers the various smart sensors and wireless systems used in these systems, their
detection mechanisms, and the advantages of contactless monitoring.

II Contactless Care Systems:

A. Contactless Care Systems for Fall Detection:

One of the primary applications of contactless care systems is the detection of falls, which
is a major concern for the elderly population, particularly those living independently. Falls
are a significant risk factor for the elderly, as they can lead to serious injuries,
hospitalization, and even mortality. Contactless fall detection systems can continuously
monitor an individual's movement and quickly identify potential falls, allowing for a rapid
response and increased safety. Ongoing research in fall detection contactless systems
utilizes a range of sensors, including cameras, radar, and motion detectors, to track an
individual's movement and identify potential falls [6]. These systems can analyze an
individual's gait, posture, and sudden movements to detect the characteristic patterns
associated with a fall [7].

B. Contactless Care Systems for Breathing Rate Monitoring:

Contactless care systems are also widely used for the continuous monitoring of breathing
rate, a vital sign that can provide valuable insights into an individual's overall health status.
These systems utilize a variety of sensors, such as cameras and radar, to track changes in an
individual's chest and abdominal movements, which are directly correlated with their
breathing patterns [4]. Traditional wearable devices for breathing rate monitoring can be
458
 Contactless Care Systems: A Review

bulky, uncomfortable, and may interfere with the individual's daily activities, leading to
poor user compliance. These limitations have hindered the widespread adoption of wearable
technology in healthcare settings, highlighting the need for more seamless and user-friendly
monitoring solutions. Contactless care systems can provide continuous, non-invasive
monitoring of breathing rate without the need for physical contact, making them a more
convenient and user-friendly solution.

III Contactless Care Systems for Heart Rate Monitoring:

In addition to fall detection and breathing rate monitoring, contactless care systems have
emerged as a promising solution for the continuous monitoring of heart rate. By leveraging
advanced sensors and signal processing algorithms, these systems can accurately track an
individual's heart rate without the need for physical contact.

A. Types of Contactless Care Systems based on Technology:

• Vision-Based Systems: Rather than relying solely on video footage, some contactless
care systems incorporate advanced computer vision algorithms and specialized cameras
to capture and analyze an individual's physical movements, breathing patterns, and even
subtle changes in skin tone that can be used to infer vital signs such as heart rate [8].
• Radar-Based Systems: Radar-based contactless care systems utilize radio waves to
detect and track an individual's movement, breathing, and heart rate.
• Motion-Based Systems: Motion-based contactless care systems employ a variety of
sensors, such as accelerometers and gyroscopes, to detect and analyze an individual's
movement and physical activity.
• Acoustic-Based Systems: Some contactless care systems utilize acoustic sensors, such
as microphones, to monitor an individual's breathing and heart rate by detecting subtle
variations in the sounds produced by the body.
• Camera-Based Systems: These systems utilize visual information captured by cameras
to monitor an individual's movements, breathing patterns, and heart rate [8].

B. Advantages of Contactless Care Systems:

Contactless care systems offer several advantages over traditional wearable devices and
manual monitoring methods, including improved user comfort and compliance, continuous
monitoring, enhanced privacy, and reduced healthcare costs [8][9].

• User Comfort and Compliance: Contactless care systems do not require the user to
wear any physical devices, eliminating the discomfort and potential skin irritation
associated with wearable sensors.
• Continuous Monitoring: Contactless care systems can provide 24/7 monitoring of an
individual's vital signs and physical activity, enabling the early detection of potential
health issues [9].
• Enhanced Privacy: Contactless care systems can monitor an individual's health
without the need for physical contact, preserving their privacy and reducing the risk of
data breaches associated with wearable devices [10].

459
International Journal of Research and Analysis in Science and Engineering

• Reduced Healthcare Costs: By enabling continuous, remote monitoring, contactless

care systems can help reduce the need for frequent in-person visits and hospital stays,
leading to decreased healthcare costs.

C. Disadvantages and Limitations:

While contactless care systems offer many advantages, they also have some limitations,
such as accuracy, interference, and limited coverage.

• Accuracy: Contactless systems may not always achieve the same level of precision as
wearable devices or manual measurements, particularly in challenging environments or
situations where the individual's movement is erratic [1].
• Interference: Factors such as environmental conditions, ambient noise, and
interference from other electronic devices can impact the performance of contactless
care systems, leading to inconsistent or unreliable data.
• Limited Coverage: Contactless care systems are typically designed to monitor
individuals within a specific area, such as a room or a designated living space, limiting
their ability to provide continuous monitoring during outdoor activities or when the
individual is away from the monitor [1] [11] [12].

IV Conclusion:

In conclusion, contactless care systems have emerged as a promising solution for the
monitoring of vital signs and the detection of abnormalities, such as falls, without the need
for physical contact. These systems utilize a variety of sensors, including cameras, radar,
and motion detectors, to track an individual's movement, breathing patterns, and heart rate,
providing a comprehensive and user-friendly approach to healthcare monitoring. The
advantages of contactless care systems, such as improved user comfort, continuous
monitoring, enhanced privacy, and reduced healthcare costs, make them a valuable tool in
the ongoing efforts to address the challenges faced by the aging population and the
increasing demand for accessible and affordable healthcare solutions.

The review of contactless care systems has highlighted the significant advancements in this
field, particularly in the areas of fall detection, breathing rate monitoring, and heart rate
monitoring. Contactless care systems offer a range of advantages over traditional wearable
devices, including improved user comfort, increased data privacy, and continuous
monitoring capabilities. As the healthcare industry continues to seek innovative solutions
to address the challenges posed by the aging population and rising costs, contactless care
systems are poised to play an increasingly crucial role in the delivery of personalized,
proactive, and cost-effective healthcare.

References:

1. W. An et al., "Smart Sensor Systems for Wearable Electronic Devices".

2. Pantelopoulos and N. Bourbakis, "A Survey on Wearable Sensor-Based Systems for
Health Monitoring and Prognosis".
3. R. Bloss, "Wearable sensors bring new benefits to continuous medical monitoring, real
time physical activity assessment, baby monitoring and industrial applications".
460
 Contactless Care Systems: A Review

4. R. D. Fazio, V. Mastronardi, M. D. Vittorio and P. Visconti, "Wearable Sensors and

Smart Devices to Monitor Rehabilitation Parameters and Sports Performance: An
Overview".
5. Zhao, M. Li and J. Z. Tsien, "The Emerging Wearable Solutions in mHealth".
6. L. N. V. Colon, Y. Delahoz and M. A. Labrador, "Human fall detection with
smartphones".
7. S. Shokri, S. Ward, P. M. Anton, P. Siffredi and G. Papetti, "Recent Advances in
Wearable Sensors with Application in Rehabilitation Motion Analysis".
8. Shao, C. Liu and F. Tsow, "Noncontact Physiological Measurement Using a Camera:
A Technical Review and Future Directions".
9. W. Hu et al., "An intelligent non-contact wireless monitoring system for vital signs and
motion detection".
10. Lo, B. Lo and G. Yang, "Transforming Health Care: Body Sensor Networks,
Wearables, and the Internet of Things".
11. M. Varma et al., "Contactless monitoring of respiratory rate (RR) and heart rate (HR)
in non-acuity settings: a clinical validity study".
12. O. H. Lowry, N. Rose rough, A. Farr and R. J. Randall, "Protein Measurement with the
Fooling Phenol Reagent".

461
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

49. Design and Modelling of Droop Control for

Small Wind Turbine Generator in µGrid
Sangamesh Y. Goudappanavar, Rubinabegaum H. Yadahalli,
Megha Sunkad, Prema, Ravindranath
Department of Electrical and Electronics Engineering,
Basaveshwar Engineering College,
Bagalkote, Karnataka State, India.

ABSTRACT:

Energy management through control strategies plays a significant role for efficient
operation of micro grid. Most of the control techniques concentrate on generation of gate
signals to operate power electronics devices. Energy management and load sharing of
various sources according to variable loads is done through power electronic devices. Also,
renewable sources mainly small wind turbine generator (SWT) is always intermittent in
nature and wind speed in some of the location is lesser. Further, efficient utilization of
energy available from SWT is done from conventional control strategy. Droop technique
in dc micro grid provides droop setting so power supplied from sources is proportional to
total load. In this project mathematical modelling of droop control system for SWT in micro
grid is implemented, including two sources and load using MATLAB/SIMULINK. To obtain
linear droop relation between small wind turbine and varying the load reference current is
modelled with respect to bus voltage. The droop control applying in micro grid system is
more efficient and increases the reliability of the system.

KEYWORDS:

Microgrid, droop control, Small Wind Turbine generator, battery, loads,

MATLAB/simulation.

I Introduction:

Micro grid (µGrid) is low voltage power grid that can operate independently with other low
voltage distribution system. Droop is common working method used in power system to
sharing the load between the various resources. Basically this is load sharing, generating
units running in parallel, for multiple generators on the same electrical grid. Droop control
is some kind of control which senses the current drawn by the load, because this current
would tend to pull down the level of the output voltage, increases some operating parameter
in the generator such as the field current, output voltage.
462
 Design and Modelling of Droop Control for Small Wind Turbine Generator in µGrid

The concept of droop control is the value of the actual DC outcome voltage linearly reduces
with increasing DC outcome energy for each system. Micro grids are power distribution
system including loads, small wind turbine, Diesel Generator (DG), battery with utility
system. It includes various low rated power generating sources that makes it highly flexible
and efficient. It allows more efficient and efficient battery, so produced power need not to
have to transfer long ranges along transmitting power for customers to receive. The
literature on control and operation of micro grids, particularly DC micro grids, offers diverse
approaches and innovations aimed at improving system stability, efficiency, and reliability.
However, several lacunae exist in the current body of research that future studies could
address. Bunker and Weaver (2014) presented an optimal geometric control strategy for DC
micro grids, focusing on precise control methods to enhance operational performance.
However, their study lacks consideration of real-world implementation challenges and
scalability of the proposed control technique across larger, more complex µGrid systems
[1]. Goudappanavar, Raghuram, and Jangamshetti (2013) proposed a simple control
strategy for a UPF power conditioner connected to a wind turbine, emphasizing effective
integration of renewable energy sources. While the control strategy is straightforward, the
study does not address the impact of varying wind conditions on system performance and
the potential need for adaptive control mechanisms [2]. Xu and Chen (2011) explored the
control and process of DC µGrids with different generation and energy storage, providing
insights into managing fluctuations in power supply. However, their work does not fully
explore the economic implications of the proposed system and the cost-benefit analysis of
integrating such control mechanisms in different grid scenarios [3]. Atur, Goudappanavar,
and Jangamshetti (2017) introduced a novel control algorithm for storage systems in small
wind turbine generators, highlighting the importance of efficient energy storage solutions.
This study, however, does not consider the long-term durability and maintenance
requirements of the storage systems, which are critical for practical deployment [4].

Chung et al. (2010) explored different control strategies for distributed generators in micro
grids that are interfaced with inverters., contributing to the optimization of power
distribution in interconnected systems. The study, though comprehensive, overlooks the
potential cybersecurity threats to inverter-interfaced systems and the need for robust
protective measures against such threats [5]. De Brabandere et al. (2007) created a technique
for controlling the voltage and frequency droop in parallel inverters., a foundational
technique for managing multiple inverters in micro grid environments. Despite its
importance, the paper does not address the limitations of droop control under highly variable
load conditions and the potential need for supplementary control strategies [6].

Prabhakaran, Goyal, and Agarwal (2018) proposed non-linear droop control methods to
address voltage regulation and issues in DC µGrids, offering solutions for improved system
stability. However, their research does not sufficiently explore the real-time performance of
these techniques under different load profiles and grid conditions [7]. Wei et al. (2015)
examined the impact of line resistance on load sharing and proposed an improved droop
control method for DC micro grids, further refining control strategies for effective power
distribution. The study, however, lacks a detailed analysis of the effects of long-distance
power transmission on load sharing and overall system efficiency [8]. Goudappanavar and
Vijayalaxmi (2024) designed an SVC for dynamic compensation to improve LVRT in
SWTG, enhancing the robustness of wind energy systems. Nonetheless, the paper does not

463
International Journal of Research and Analysis in Science and Engineering

address the cost-effectiveness and potential economic barriers to implementing SVC

systems in small-scale wind turbines [9].

Savaghebi et al. (2013) centered on droop-controlled, autonomous voltage imbalance

adjustment in islanded micro grids, presenting a method for maintaining voltage stability in
isolated systems. While effective, the study does not consider the integration of this
compensation method with other existing control systems in micro grids, potentially
limiting its applicability [10].

Katiraei, Iravani, and Lehn (2005) investigated micro grid autonomous operation during
and after islanding, providing crucial insights into micro grid resilience. However, their
work does not sufficiently address the coordination challenges between multiple micro grids
or the scalability of the proposed methods [11]. Salomonsson, Soder, and Sannino (2009)
discussed protection strategies for low-voltage DC micro grids, emphasizing the need for
robust protective measures. Despite its thoroughness, the study overlooks the impact of
emerging technologies, such as advanced sensors and AI-based protection systems, on the
effectiveness of protection strategies [12].

Goudappanavar and Jangamshetti (2020) conducted stochastic power flow analysis of

unbalanced distribution systems using clusters of small wind turbine generators and solar
PV systems, contributing to the understanding of power flow dynamics under uncertainty.
The study, however, does not explore the integration of stochastic analysis with real-time
monitoring systems for enhanced operational decision-making [13].

Chen et al. (2015) introduced a nonlinear droop method to improve voltage regulation and
load sharing in DC systems, advancing control strategies for better system performance.
Nonetheless, the research lacks a comprehensive evaluation of the long-term stability and
resilience of the nonlinear droop method under varying operational conditions [14].
Guerrero et al. (2011) proposed a hierarchical control approach for droop-controlled AC
and DC micro grids, aiming at standardization and improved control architecture. While
significant, the study does not address the interoperability issues between different
hierarchical control systems and the potential need for global standards [15].

Lastly, Goudappanavar, Jangamshetti, and O (2021) developed a method for identifying

fault areas in distribution systems based on the rotor angle behavior of diesel generators,
enhancing fault detection and localization capabilities in power distribution networks.
Nevertheless, the impact of incorporating renewable energy sources is not taken into
account in this work. on the proposed fault identification method, which could affect its
effectiveness in modern power systems [16].

Category-2 information required for droop control system with micro grid modelled, WTG,
Energy storage device, loads equations of the proposed system. Category-3 includes
simulation model of proposed system for constant and variable loads. Category-4 provides
the simulation results and discussion. Category-5 includes conclusion and future scope.

464
 Design and Modelling of Droop Control for Small Wind Turbine Generator in µGrid

II Mathematical Modelling of Micro Grid:

The droop control numerical approach is modelled and developed using MATLAB/
SIMULINK.

Figure 1: Block Diagram of Micro Grid System

Figure 1 shows block diagram of micro grid system, including small wind turbine, Battery
and various loads. In the proposed system, a buck-boost converter based circuit topology is
considered for analysis. In proposed system, a buck boost converter based circuit topology
is considered for analysis. In this system the small wind turbine is DC source is connected
to the bus through buck boost converter it converts the DC to DC voltage to get the load.
The bus voltage is DC voltage. Load is consisting of R-C load. Battery is storage device in
this system battery is considered as lead-acid battery. It is also connected to the bus through
buck boost converter for step down the voltage.

Figure 2: Simple Micro Grid System

465
International Journal of Research and Analysis in Science and Engineering

Figure 2 shows circuit of micro grid for this is design based droop control. The simple small
micro grid example with two resources and one load. Both resources and load linked with
the common bus. In this micro grid system battery is act as source. When the load differs
causes modify the bus voltage. Source one is small wind turbine is connected to the bus,
source two is battery connected to the common bus and R-C load is connected to same bus.
Here bus is common for both sources and load.

Applying KVL (Kirchhoff’s Voltage Law), and KCL (Kirchhoff’s Current Law) on both
sources and load. Current equation [1] for small wind turbine, the air particles having mass
m and velocity v, the kinetic energy carried by the wind particles by the equation:

1 (1)
𝐸𝐾𝑖𝑛 = 𝑚𝑣 2
2
𝑚 = 𝜌𝜋𝑟 2 vt (2)
1 (3)
𝐸𝐾𝑖𝑛 = 𝜌𝜋𝑟 2 𝑣 3 𝑡
2

Wind power at time as,

1 (4)
𝑃𝑊𝑖𝑛𝑑 = 𝜌𝜋𝑟 2 𝑣 3
2

Charging voltage from source,

𝑑𝑉𝑐ℎ 𝑉𝑟𝑒𝑓 − 𝑉𝑏𝑢𝑠 𝑉𝑤 (5)

𝐶ℎ = 𝑖𝑙ℎ − (𝐾𝑝𝑣 [𝐾𝑝𝑖 ( + − 𝑖𝑜𝑢𝑡 ) + 𝐾𝑖𝑖 𝑒𝑟𝑟𝑜𝑟1 − 𝑉𝑐𝑙 ]
𝑑𝑡 𝑅𝑑1 𝑅𝑑2

+𝐾𝑖𝑣 𝑒𝑟𝑟𝑜𝑟2 )𝑖𝑙𝑙

The reference current of source is

𝑉𝑟𝑒𝑓 − 𝑉𝑏𝑢𝑠 𝑉𝑤 (6)

𝑖𝑟𝑒𝑓 = +
𝑅𝑑1 𝑅𝑑2

Here reference current improves with wind speed.

𝑑𝑉𝑐ℎ (7)
𝐶 = 𝑖𝑙ℎ − (𝐾𝑝𝑣 [𝐾𝑝𝑖 (𝑖𝑟𝑒𝑓 ) + 𝐾𝑖𝑖 𝑒𝑟𝑟𝑜𝑟1 − 𝑉𝑐𝑙 ] + 𝐾𝑖𝑣 𝑒𝑟𝑟𝑜𝑟2 )𝑖𝑙𝑙
𝑑𝑡 ℎ

Output current 𝑖𝑜𝑢𝑡 is,

466
 Design and Modelling of Droop Control for Small Wind Turbine Generator in µGrid

𝑑𝑖𝑜𝑢𝑡 (8)
𝐿 = 𝑉𝑐𝑙 − 𝑅𝐵 𝑖𝑜𝑢𝑡 − 𝑉𝑏𝑢𝑠
𝑑𝑡 𝐵

Constant error from source,

𝑑𝑒𝑟𝑟𝑜𝑟1 𝑉𝑟𝑒𝑓 − 𝑉𝑏𝑢𝑠 𝑉𝑊 (9)

= + − 𝑖𝑜𝑢𝑡
𝑑𝑡 𝑅𝑑1 𝑅𝑑2
𝑑𝑒𝑟𝑟𝑜𝑟2 (10)
= 𝐾𝑝𝑖 (𝑖𝑟𝑒𝑓 − 𝑖𝑜𝑢𝑡 ) + 𝐾𝑖𝑖 𝑒𝑟𝑟𝑜𝑟1 − 𝑉𝑐𝑙
𝑑𝑡

Battery, charging voltage equation [1] is

𝑑𝑉𝑐ℎ (11)
𝐶 = 𝑖𝑙ℎ − 𝐷𝑖𝑙𝑙
𝑑𝑡 ℎ

Output current is,

𝑑𝑖𝑜𝑢𝑡 (12)
𝐿 = 𝑣𝑐𝑙 − 𝑅𝐵 𝑖𝑜𝑢𝑡 − 𝑉𝑏𝑢𝑠
𝑑𝑡 𝐵

Battery voltage is,

𝑑𝑉𝑐𝑏𝑎𝑡𝑡 (13)
𝐶𝑏𝑎𝑡𝑡 = 𝑖𝑙ℎ
𝑑𝑡

Loads, Variable and constant loads are two types of loads. The load voltage equation is,

𝑑𝑉𝑐ℎ (14)
𝐶 = 𝑖𝑖𝑛 − 𝐷𝑖𝑙𝑙
𝑑𝑡 ℎ

Apply P-I controller, the constant error is,

𝐷 = 𝐾𝑝𝑙 (𝑉𝑛𝑜𝑚 − 𝑉𝑐𝑙 ) + 𝐾𝑖𝑙 𝑒𝑟𝑟𝑜𝑟 (15)

Where D is duty cycle changed to P-I Controller,

For complete micro grid, Bus voltage is common for both sources and load,

467
International Journal of Research and Analysis in Science and Engineering

𝑑𝑉𝑏𝑢𝑠 𝑉𝑏𝑢𝑠 (16)

𝐶𝑙𝑜𝑎𝑑 = 𝑖1 − 𝑖2 −
𝑑𝑡 𝑅𝑙𝑜𝑎𝑑

Control P-I loop for source one (small wind turbine),

𝑉1𝑆 = 𝐾𝑝 (𝑖𝑟𝑒𝑓1 − 𝑖1 ) + 𝐾𝑖 𝑖𝑒𝑟𝑟𝑜𝑟,1 (17)

For source two (Battery) the control P-I loop is,

𝑉2𝑆 = 𝐾𝑝 (𝑖𝑟𝑒𝑓2 − 𝑖2 ) + 𝐾𝑖 𝑖𝑒𝑟𝑟𝑜𝑟,2 (18)

The reference current for source two conventional straight line droop control will be
implemented,

−𝑉𝑏𝑢𝑠 + √4𝑃𝑅1𝐵 + 𝑉 2 𝑏𝑢𝑠 (19)

𝑖 ∗ 𝑟𝑒𝑓1 =
2𝑅1𝐵

III Modelling and Simulation of Proposed System:

A. Modelling of µGrid in MATLAB/Simulink: The system is simulated using droop

management strategy and modelled in MATLAB/SIMULINK as shown in Figure 3.1
Further, droop control connection between bus voltages and reference current to
complete given object.

Figure 3: Simulation Model of Micro grid

Figure 3 represents the simulation model of micro grid system containing small wind
turbine, battery, constant load and variable loads are modeled. In this model all the sources
468
 Design and Modelling of Droop Control for Small Wind Turbine Generator in µGrid

are combined from equations [1]. Mathematical model is consisting of four subsystems are
small wind turbine generator, battery, constant and variable loads. Each system modelled
by mathematical equations [1] shown in above figure. 1.5 kW wind turbine powered
generator is modelled in MATLAB using equations [1], which research has been taken with
continuous wind speed of 8 m/s and varying load from wind turbine at 20m at BEC
Bagalkot.

B. µGrid for Variable Load:

In this micro grid system variable loads are considered. Change in load simulation also
implemented. Figure 3.5 power curve graph shows the load data with time for every 1 hr.,
taken from BEC SCADA for Distribution and Automation Research Center on 5 th July
2018.

Figure 4: Shows The Power Curve for Variable Load.

IV Simulation Results:

The suggested droop management strategy, the µGrid is modelled applying

MATLAB/SIMULINK. The system is simulated using straight line droop management for
both sources droop control relationships. The wind powered generator and battery same
straight line droop management for both models, Using maximum droop management
relationship for both sources. For this, a simulation model of the µGrid is implemented.
Power electronics (Buck boost converter) connected source to the bus. The droop control
relationship constant output of 2.1 kW from small wind turbine.

469
International Journal of Research and Analysis in Science and Engineering

Figure 5: Bus Voltage of Proposed Micro grid System

Figure 6: Relationship of Droop control of micro grid System for WTG

Figure 7: Optimal Power Provided by WTG and Battery

470
 Design and Modelling of Droop Control for Small Wind Turbine Generator in µGrid

Figure 8: Output Voltage for WTG

Case Study:

A. Micro grid control for variable Load:

Micro grid considered above is carried out for variable loads taken from BEC SCADA on
5th July 2018.

Figure 9: Power Output with of WTG Corresponding to Variable Load

471
International Journal of Research and Analysis in Science and Engineering

Figure 10: Power Provided with Load By Battery

V Conclusion:

µGrid model is simulated using MATLAB/SIMULINK. The simulation complete design of

the µGrid is applied such as power electronic devices elements that are linked source to bus.
The simulation of complete design of the micro grid is applied such as power electronic
devices elements that are linked source to bus. The goal is to keep the source's output power
constant. It was used once more with wind resources, where the load and WTG determine
the desired output power. The proposed control method is implemented. Two sources and a
battery are included in a grid simulation that uses optimal droop management. Using
measured load data, simulation results were shown to illustrate how the system operated
over the course of a day. Then, the bus voltage and wind speed determine the droop control
reference current.

Implement droop control for wind energy conversion system, and also find the droop
relationship to optimize the source operation. The work presented in this project method for
to obtain linear droop relation between small wind turbine and varying load. with simulation
results would certainly help in improving voltage stability. The control technique does not
need connections between system elements.

The implementing of additional features preferred power from source is a function of

another photo voltaic (PV) system, fuel cell with using small wind turbine generator.

Acknowledgment:

The authors thank the TEQIP-II, Basaveshwar Engineering College Bagalkote, Karnataka
State for financial assistance in the seed grant scheme to setup and utilize the research
facility.

References:

1. K. J. Bunker and W. W. Weaver, "Optimal geometric control of Dc micro grids," 2014

IEEE 15th Workshop on Control and Modeling for Power Electronics (COMPEL),
Santander, Spain, 2014, pp. 1-6, doi: 10.1109/COMPEL.2014.6877217.
472
 Design and Modelling of Droop Control for Small Wind Turbine Generator in µGrid

2. Sangamesh Y. Goudappanavar, Naik Raghuram L., Jangamshetti Suresh H., “Simple

Control Strategy for UPF Power Conditioner of GSC connected to Wind Turbine”,
IASTED Asia PES-2013, April 10-12, 2013, Phuket Thailand.
DOI: 10.2316/P.2013.800-060.
3. L. Xu and D. Chen, "Control and Operation of a DC Micro grid with Variable
Generation and Energy Storage," in IEEE Transactions on Power Delivery, vol. 26, no.
4, pp. 2513-2522, Oct. 2011, doi: 10.1109/TPWRD.2011.2158456.
4. V. S. Atur, S. Y. Goudappanavar and S. H. Jangamshetti, "Novel Control Algorithm of
Storage System for Small Wind Turbine Generator," 2017 International Conference on
Current Trends in Computer, Electrical, Electronics and Communication (CTCEEC),
Mysore, India, 2017, pp. 383-388, doi: 10.1109/CTCEEC.2017.8455184.
5. Y. Chung, W. Liu, D. A. Cartes, E. G. Collins and S. -I. Moon, "Control Methods of
Inverter-Interfaced Distributed Generators in a Micro Grid System," in IEEE
Transactions on Industry Applications, vol. 46, no. 3, pp. 1078-1088, May-June 2010,
doi: 10.1109/TIA.2010.2044970.
6. K. De Brabandere, B. Bolsens, J. Van den Keybus, A. Woyte, J. Driesen and R.
Belmans, "A Voltage and Frequency Droop Control Method for Parallel Inverters," in
IEEE Transactions on Power Electronics, vol. 22, no. 4, pp. 1107-1115, July 2007, doi:
10.1109/TPEL.2007.900456.
7. P. Prabhakaran, Y. Goyal and V. Agarwal, "Novel Nonlinear Droop Control
Techniques to Overcome the Load Sharing and Voltage Regulation Issues in DC Micro
grid," in IEEE Transactions on Power Electronics, vol. 33, no. 5, pp. 4477-4487, May
2018, doi: 10.1109/TPEL.2017.2723045.
8. C. Wei, Y. Daopei, C. Jialin, W. Yongzhi and Y. Yuxing, "The impact of line resistance
on load sharing and an improved droop control of DC micro grid," 2015 9th
International Conference on Power Electronics and ECCE Asia (ICPE-ECCE Asia),
Seoul, Korea (South), 2015, pp. 208-212, doi: 10.1109/ICPE.2015.7167788.
9. Sangamesh Y. Goudappanavar, Vijayalaxmi R. H., "Design of SVC for Dynamic
Compensation to Improve LVRT of Small Wind Turbine", IJIREEICE International
Journal of Innovative Research in Electrical, Electronics, Instrumentation and Control
Engineering, vol. 12, no. 4, 2024, Crossref
https://doi.org/: 10.17148/IJIREEICE.2024.12403.
10. M. Savaghebi, A. Jalilian, J. C. Vasquez and J. M. Guerrero, "Autonomous Voltage
Unbalance Compensation in an Islanded Droop-Controlled Micro grid," in IEEE
Transactions on Industrial Electronics, vol. 60, no. 4, pp. 1390-1402, April 2013, doi:
10.1109/TIE.2012.2185914.
11. F. Katiraei, M. R. Iravani and P. W. Lehn, "Micro-grid autonomous operation during
and subsequent to islanding process," in IEEE Transactions on Power Delivery, vol. 20,
no. 1, pp. 248-257, Jan. 2005, doi: 10.1109/TPWRD.2004.835051.
12. D. Salomonsson, L. Soder and A. Sannino, "Protection of Low-Voltage DC Micro
grids," in IEEE Transactions on Power Delivery, vol. 24, no. 3, pp. 1045-1053, July
2009, doi: 10.1109/TPWRD.2009.2016622.
13. S. Y. Goudappanavar and S. H. Jangamshetti, "Stochastic Power Flow Analysis of
Unbalanced Distribution System based on cluster of SWTG-SPV," 2020 2nd PhD
Colloquium on Ethically Driven Innovation and Technology for Society (PhD EDITS),
Bangalore, India, 2020, pp. 1-2, doi: 10.1109/PhDEDITS51180.2020.9315298.
14. F. Chen, R. Burgos, D. Boroyevich and W. Zhang, "A nonlinear droop method to
improve voltage regulation and load sharing in DC systems," 2015 IEEE First

473
International Journal of Research and Analysis in Science and Engineering

International Conference on DC Micro grids (ICDCM), Atlanta, GA, USA, 2015, pp.
45-50, doi: 10.1109/ICDCM.2015.7152008.
15. J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuna and M. Castilla, "Hierarchical
Control of Droop-Controlled AC and DC Micro grids—A General Approach Toward
Standardization," in IEEE Transactions on Industrial Electronics, vol. 58, no. 1, pp. 158-
172, Jan. 2011, doi: 10.1109/TIE.2010.2066534.
16. S. Y. Goudappanavar, S. H. Jangamshetti and J. J. O, "Identification of Fault Area in
the Distribution System based on Rotor Angle behaviour of Diesel Generator," 2020
3rd International Conference on Energy, Power and Environment: Towards Clean
Energy Technologies, Shillong, Meghalaya, India, 2021, pp. 1-6,
doi: 10.1109/ICEPE50861.2021.9404451.

474
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

50. LoRa based Smart Energy Meter for Theft

Detection
Akash P. S., Deepa S. Patil, Sagar N. Odunavar,
Nagaprasad H.
Department of Electrical and Electronics Engineering,
Basaveshwar Engineering College, Bagalkote, Karnataka, India.
Chayalakshmi C. L.
Associate Professor, Department of Electrical and Electronics Engineering
Basaveshwar Engineering College, Bagalkote, Karnataka, India.

ABSTRACT:

LoRa (Long Range) is a wireless communication technology designed for long-range, low-
power, low-data-rate applications. It is commonly used in Internet of Things (IoT) devices
due to its ability to communicate over distances of several kilometers while consuming
minimal power. LoRa technology is ideal for creating efficient, low-power, and long-range
communication networks, especially for IoT applications. It enables the deployment of a
wide variety of smart applications, from smart cities to industrial monitoring, making it a
key technology in the advancement of the IoT ecosystem. The growing need for efficient and
secure energy management systems has led to the development of smart energy meters. This
paper presents a novel LoRa-based smart energy meter designed to detect and prevent
electricity theft. Utilizing the LoRa (Long Range) communication technology, the proposed
system offers a reliable, long-range, and low-power solution for real-time monitoring of
energy consumption and theft detection. The smart energy meter integrates a digital energy
meter with a microcontroller and LoRa module, which transmits data to a central server.
The server processes the data using advanced algorithms to identify irregular consumption
patterns indicative of theft. This system enhances the traditional energy meter by providing
continuous, wireless data transmission over long distances, making it suitable for
deployment in both urban and rural areas. This paper presents a LoRa-based smart energy
meter designed for theft detection and remote meter reading. The proposed LoRa-based
smart energy meter offers significant benefits, including improved theft detection accuracy,
reduced power consumption, and extended communication range. This makes it a cost-
effective and scalable solution for energy providers seeking to enhance the security and
efficiency of their power distribution networks. The implementation details, including the
hardware setup, and software algorithms are discussed to provide a comprehensive
overview of the system's functionality and performance.

475
International Journal of Research and Analysis in Science and Engineering

KEYWORDS:

LoRa, Smart Energy Meter, Electricity Theft Detection, Long-Range Communication, Low
Power Consumption, Real-Time Monitoring, Energy Management, Anomaly Detection,
Wireless Data Transmission.

I Introduction:

The advent of smart technologies has revolutionized various sectors, including energy
management. Traditional energy meters, while essential for measuring consumption, often
fall short in terms of real-time monitoring and theft detection. Electricity theft remains a
significant challenge for energy providers worldwide, leading to substantial economic
losses and operational inefficiencies. In this context, the integration of advanced
communication technologies, such as LoRa (Long Range), with smart energy meters
presents a promising solution.

LoRa is a wireless communication technology known for its long-range capabilities and low
power consumption, making it ideal for Internet of Things (IoT) applications. It operates in
the unlicensed ISM bands, providing a cost-effective means of communication over large
distances. By leveraging LoRa technology, smart energy meters can continuously monitor
and transmit data on energy consumption, enabling real-time detection of anomalies and
potential theft.

The deployment of smart meters equipped with Low-Power Wide-Area Network (LPWAN)
technologies, particularly LoRa (Long Range), is revolutionizing utility infrastructure by
enhancing data collection, transmission, and analysis. LoRa technology offers a unique
combination of long-range transmission capabilities and low power consumption, making
it ideal for extensive utility networks. This technology's robust performance in challenging
environments and its ability to maintain stable communication over vast distances
contribute to reduced implementation costs and increased network reliability. By integrating
LoRa-enabled smart modules into energy meters, utility providers can efficiently monitor
and manage utility usage, benefiting from real-time insights and operational efficiencies.

One of the key advantages of LoRa technology is its programmability, which allows for
scheduled notifications and updates, thereby enhancing the responsiveness of utility
management systems. This feature enables timely interventions and accurate meter
readings, consumption alerts, and system diagnostics, providing both providers and
consumers with actionable data. Compared to traditional communication technologies like
Zigbee, Wi-Fi, and Bluetooth, which are limited by their range and power requirements,
LoRa excels in providing long-range, energy-efficient connectivity. This makes it
particularly suitable for applications in smart cities, smart grids, and other large-scale IoT
implementations where reliable, cost-effective communication is essential.

LoRa's use of a patented chirped spread spectrum modulation technique distinguishes it

from other LPWAN technologies like Sig fox, enhancing data transmission reliability even
in congested environments. The LoRa WAN protocol, built on LoRa technology, further
optimizes network performance with adaptive rate capabilities.

476
 LoRa based Smart Energy Meter for Theft Detection

LoRa networks typically utilize a star topology, where smart meters communicate directly
with gateways that relay data to central servers via standard IP protocols. This architecture
simplifies deployment and reduces costs, making LoRa a compelling choice for utility
providers seeking to modernize their infrastructure efficiently. Overall, LoRa technology is
driving a significant transformation in utility management, offering a scalable, cost-
effective solution that meets the evolving demands of an interconnected world.

Motivation:

Communication channels such as Wi-Fi, Zigbee, and Bluetooth require high power and
have limited range. However, Low Power Wide Area Network (LPWAN) technology
utilizes minimal power and enables data transmission without internet connectivity.
Existing energy meters provide real-time data on energy consumption, voltage, and current,
but they lack the capability to track previous months' energy usage. Consequently, meter
readers visit each household at the beginning of every month to deliver energy bills
Implementing a system for transmitting energy consumption data from existing meters to a
central substation can eliminate the need for meter readers and facilitate the generation of
bills directly on a website hosted by the substation. This data can be stored both at the
consumer's premises and in the substation. This approach aids in load estimation on the
distribution side and enhances demand-side management. By comparing the energy
consumption of individual consumers with the aggregate consumption at the distribution
level, instances of energy theft can be identified. If the total energy consumption of all
homes matches the reading from the distribution energy meter, accounting for some
percentage of transmission losses, then no theft has occurred. However, any disparities
indicate potential energy theft.

Objectives:

The primary objective of this research is to develop a LoRa-based smart energy meter
system that enhances the accuracy and reliability of electricity theft detection. The proposed
system combines a digital energy meter with a microcontroller and a LoRa module to
facilitate long-range wireless communication. Data collected from the energy meter and
sensors are transmitted to a central server, where advanced algorithms analyze the data for
irregularities indicative of theft.

This system addresses the limitations of conventional energy meters by providing

continuous monitoring and immediate alerts in the event of suspicious activity. The use of
LoRa technology not only extends the communication range but also ensures low power
consumption, making it suitable for deployment in both urban and rural areas. Moreover,
the scalability of the system allows for widespread adoption, providing energy providers
with a robust tool for improving the security and efficiency of their power distribution
networks.

This paper discusses the design and implementation of the LoRa-based smart energy meter
system, including the hardware components, software algorithms, and communication
protocols. The potential benefits, such as enhanced theft detection accuracy, reduced
operational costs, and improved energy management, are also highlighted.

477
International Journal of Research and Analysis in Science and Engineering

Through this research, we aim to demonstrate the effectiveness of integrating LoRa

technology with smart energy meters in mitigating electricity theft and fostering more
efficient energy utilization.

II Literature Review:

LoRa (Long Range) is a low-power, wide-area network (LPWAN) technology that supports
long-range communication, making it ideal for IoT applications in various sectors, including
energy management. LoRa operates in the unlicensed ISM bands, offering a cost-effective
solution for long-distance data transmission. Key features of LoRa include its ability to
provide robust communication over several kilometers, low power consumption, and
support for a large number of devices in a single network. These characteristics make LoRa
particularly suitable for applications requiring reliable, long-range connectivity and
extended battery life. this section provides the overall literature available at present.

The paper "Wireless Data Transmission Using LoRa" by H. Venkatesh, M. Lakshmanna,

and B. Mansoor discusses the implementation and advantages of using LoRa (Long Range)
technology for wireless data transmission. The authors focus on the technical aspects,
performance benefits, and potential applications of LoRa in various fields. The paper
provides an overview of LoRa technology, highlighting its ability to support long-range
communication with low power consumption. LoRa operates in the unlicensed ISM bands,
making it a cost-effective solution for wireless communication. The authors delve into the
technical specifications of LoRa, including its modulation technique (Chirp Spread
Spectrum), which allows for robust communication over long distances. The paper explains
how LoRa can achieve long-range communication (up to 15 kilometers in rural areas) while
maintaining low power usage, making it suitable for battery-powered devices. A practical
implementation of a LoRa-based wireless data transmission system is presented. The
authors describe the setup, which includes LoRa modules, microcontrollers, and the
necessary interfacing components. Details on the coding and configuration required for
establishing communication between the devices are provided. The paper evaluates the
performance of the LoRa system in various environments, such as urban and rural settings.
Metrics such as range, data rate, and power consumption are analyzed to demonstrate the
effectiveness of LoRa technology. The authors conclude that LoRa technology offers a
promising solution for long-range, low-power wireless communication. The paper suggests
that further research and development could enhance the capabilities and applications of
LoRa, making it an integral part of the growing IoT ecosystem [1].

The paper "IoT Based Smart Energy Meter for Efficient Energy Utilization in Smart Grid"
by B. K. Barman, S. N. Yadav, S. Kumar, and S. Gope explores the development and
implementation of a smart energy meter leveraging IoT (Internet of Things) technology.
The focus is on enhancing energy utilization and efficiency within the framework of a smart
grid. The paper introduces the concept of smart grids and the role of IoT in modernizing
energy management systems. Emphasis is placed on the need for efficient energy utilization
to reduce wastage and improve the reliability of power distribution. The authors describe
the architecture of the IoT-based smart energy meter, which includes components such as
microcontrollers, communication modules, and sensors. The smart meter is designed to
measure and monitor electricity consumption in real-time, providing detailed usage data to
both consumers and utility providers.

478
 LoRa based Smart Energy Meter for Theft Detection

The paper highlights the integration of IoT technology, enabling the smart meter to transmit
data wirelessly to a central server or cloud platform. Various communication protocols and
technologies, such as Wi-Fi, Zigbee, and cellular networks, are discussed in the context of
data transmission. The authors conclude that IoT-based smart energy meters represent a
significant advancement in energy management, contributing to the overall efficiency and
reliability of smart grids. The paper calls for further research and collaboration between
industry and academia to address the challenges and fully realize the potential of IoT in
smart energy systems [2].

The paper "Development of a Low-cost LoRa based SCADA system for Monitoring and
Supervisory Control of Small Renewable Energy Generation Systems" by C. Ndukwe, M.
T. Iqbal, and J. Khan discusses the creation and implementation of a Supervisory Control
and Data Acquisition (SCADA) system that uses LoRa (Long Range) technology.

The system is designed for the efficient monitoring and control of small-scale renewable
energy generation systems, providing a cost-effective solution suitable for remote and
resource-constrained environments. The paper begins by highlighting the importance of
monitoring and controlling renewable energy systems to ensure optimal performance and
reliability. Traditional SCADA systems are often expensive and complex, posing challenges
for small-scale and remote renewable energy projects.

The authors propose a low-cost alternative using LoRa technology. The paper provides an
overview of LoRa technology, emphasizing its long-range communication capabilities, low
power consumption, and suitability for remote monitoring applications. LoRa's ability to
operate in the unlicensed ISM bands makes it a cost-effective choice for implementing
wireless SCADA systems. The authors describe the implementation process, including the
setup of LoRa communication modules, integration with sensors for monitoring parameters
such as voltage, current, and temperature, and the configuration of microcontrollers for data
acquisition. Data is transmitted wirelessly to a central server where it is processed and
stored. The paper concludes that the development of a low-cost LoRa-based SCADA system
offers a practical and effective solution for monitoring and controlling small renewable
energy generation systems. The proposed system addresses the limitations of traditional
SCADA systems, making advanced monitoring and control accessible to a wider range of
renewable energy projects [3].

The paper "Smart Energy Metering and Power Theft Control Using Arduino & GSM" by
A. S. Metering, S. Visalatchi, and K. K. Sandeep explores the design and implementation
of a smart energy metering system aimed at monitoring electricity usage and detecting
power theft. The system utilizes Arduino microcontrollers and GSM (Global System for
Mobile Communications) technology to provide real-time data transmission and alerts. The
paper introduces the problem of power theft and its significant impact on the energy sector,
including financial losses and reduced system efficiency. The authors propose a smart
energy metering system that leverages modern technologies to enhance the accuracy of
energy usage measurement and prevent unauthorized electricity consumption. The smart
energy meter is built around an Arduino microcontroller, which serves as the central
processing unit. Key components include current and voltage sensors for measuring power
consumption, a GSM module for wireless communication, and an LCD display for local
data visualization.
479
International Journal of Research and Analysis in Science and Engineering

The paper provides a detailed description of the hardware setup, including the interfacing
of sensors with the Arduino microcontroller and the configuration of the GSM module for
data transmission.

Software algorithms for data processing, theft detection, and communication are also
discussed. The paper concludes that the integration of Arduino and GSM technology in
smart energy meters offers a practical and efficient solution for accurate energy monitoring
and effective power theft control. The proposed system has the potential to significantly
reduce energy losses and improve the overall efficiency of power distribution networks [4].

In existing systems, either an electronic energy meter or an electro-mechanical meter is

installed on the premises for measuring consumption. These meters currently in use are only
capable of recording kWh units. The kWh units still have to be recorded monthly by meter
readers, who must walk from building to building. The recorded data needs to be processed
by a meter reading company. For processing the meter reading, the company must first link
each recorded power usage datum to an account holder and then determine the amount owed
by means of the specific tariff in use. Wireless smart energy meters are replacing traditional
meters to ensure accurate tariff calculation and reduce errors caused by human beings.
readers. These smart energy meters utilize GSM, Wi-Fi, and other wireless technologies.
The main drawback of these systems is the necessity for network access on the consumer
side for the smart energy meter to connect wirelessly.

III Proposed System:

The current system of electricity consumption billing has some errors in recording and also
is very time consuming. Errors are likely to be introduced at every stage due to electro-
mechanical meters, human errors while noting down the meter reading and errors while
processing the paid bills and the due bills.

Smart energy meter is a novel technique which can reduce these problems associated with
billing and also reduces the deployment of manpower for recording meter readings. It has
many advantages from both the distributor side as well as the consumer’s point. This smart
energy meter has been developed based on LoRa technology.

While using the LoRa technology, the disadvantages that are associated when using the
GSM, Wi-Fi like wireless networks can be overcome. It does not require any additional
towers or network access in the consumer side for these smart energy meters to connect in
wireless mode. So, these smart energy meters have the data that is transmitted wireless from
the consumer to the distributor and the Substation side using LoRa technology.

The block diagram as shown in Figure 1, consists of energy meter, LDR, Arduino, LoRa
module and display. The energy meter is connected to the transformer, and an LDR (Light
Dependent Resistor) is connected to the energy meter. This setup is linked to an Arduino,
which processes the data. The processed data is then transmitted and received via a LoRa
module. Finally, the Arduino further processes the received data and displays it.

480
 LoRa based Smart Energy Meter for Theft Detection

Figure 1: Block Diagram of the Proposed System

• In this proposed model the energy meter is connected to the low voltage (LV) side of
the transformer, while a Light Dependent Resistor (LDR) is linked to the energy meter
to detect the blinks or pulses emitted by the meter.
• An Arduino board is employed to count these pulses and subsequently transmit the data
via a LoRa (Long Range) module.
• This transmitted information is then wirelessly received by another Arduino acting as a
LoRa receiver at substation.
• This second Arduino is interfaced with a PC, facilitating data acquisition.
• The transmitted data is meticulously compared with the energy meter data of the
distribution transformer. Any disparities detected are indicative of potential energy
theft, prompting the system to issue an alert to the substation for further action.
• This setup thus ensures efficient monitoring and detection of unauthorized energy
consumption, enhancing the overall integrity of the electrical distribution system.

IV Results and Discussions:

The image depicts a setup for interfacing utility meters with a laptop using microcontrollers,
as part of a project to read meter data programmatically. The connection diagram of the
proposed model is shown in Figure 2.

481
International Journal of Research and Analysis in Science and Engineering

Figure 2: Connection Diagram of the Proposed Model

The image in Figure 3 shows a breadboard setup with an OLED display connected to a
sensor, displaying date, time, and temperature readings.

Figure 3: Display Module to Calculate Count pulses

Figure 4: Data Transmission from Sender to Receiver Module

482
 LoRa based Smart Energy Meter for Theft Detection

The image in Figure 4 shows the serial monitor data transmission from sender to receiver,
where packets of data are being received and logged with timestamps.

The LoRa-based smart energy meter system can be further enhanced by incorporating
advanced machine learning algorithms to predict and detect more sophisticated energy theft
patterns. Integration with smart grid technology can enable real-time energy management
and dynamic tariff adjustments, improving grid efficiency and consumer engagement.
Expanding the system's compatibility with various IoT platforms will facilitate seamless
data analysis and remote monitoring. Additionally, incorporating renewable energy sources
and storage solutions can optimize energy distribution and contribute to sustainable energy
management. Future iterations could also include enhanced security protocols to safeguard
data transmission and prevent cyber threats, ensuring a more robust and secure energy
distribution network.

V Conclusion:

The presented LoRa-based smart energy meter system offers a comprehensive solution for
efficient meter reading and theft detection in energy distribution networks. By seamlessly
integrating with existing infrastructure, the system minimizes disruption while enhancing
operational efficiency. Through the utilization of components like LDR sensors, Arduino
microcontrollers, and LoRa modules, the system effectively captures energy consumption
data and transmits it to the substation for billing purposes. Moreover, the incorporation of
theft detection mechanisms, including comparison of household and distribution
transformer data, enables timely identification of irregularities, enhancing security and
revenue protection for energy providers. Tested with a bulb load, the system demonstrates
promising effectiveness in detecting theft and generating accurate billing information.
Overall, the project showcases significant potential in addressing challenges related to meter
reading and theft detection, offering a scalable and reliable solution for energy
management in communities.

The proposed system builds upon the existing digital energy meter, enhancing its
capabilities without requiring the replacement of all current meters with smart energy
meters. While the installation of smart meters may generate electronic waste by rendering
existing meters obsolete, this system mitigates such waste and is more cost-effective in
comparison. It optimizes the functionality of current meters, offering an efficient and
economically viable solution for energy management.

References:

1. K. Barman, S. N. Yadav, S. Kumar, S. Gope, “IoT Based Smart Energy Meter for
Efficient Energy Utilization in Smart Grid,”2nd International Conference on Power,
Energy and Environment: Towards Smart Technology (ICEPE), Shillong, India, 2018,
pp. 1-5, doi: 10.1109/EPETSG.2018.8658501
2. C. Ndukwe, M. T. Iqbal, J. Khan, “Development of a Low-cost LoRa based SCADA
system for Monitoring and Supervisory Control of Small Renewable Energy Generation
Systems,” International Electronics and Mobile Communication Conference, Canada,
2020, pp. 0479-0484, doi: 10.1109/IEMCON51383.2020.9284933.

483
International Journal of Research and Analysis in Science and Engineering

3. A. S. Metering, S. Visalatchi, K. K. Sandeep, “Smart energy metering and power theft

control using arduino & GSM,”2nd International Conference for Convergence in
Technology (I2CT), Mumbai, India, 2017, pp. 858-961,
doi: 10.1109/I2CT.2017.8226251.
4. H Venkatesh, M Lakshmanna, B Mansoor “Wireless Data Transmission Using Lora”
IJFANS International Journal of Food and Nutritional Sciences, Vol. 11, Issue 8, 2022,
pp. 3590-3596.
5. "Wireless Sensor Network Using LoRa and Arduino" project by Hackster.io
(https://www.hackster.io/aryanshsingh/wirelesssensor-network-using-loraand-
arduino-1fdd51)
6. C. S. Choi, J. D. Jeong, I. W. Lee, W. K. Park, “Lora Based Renewable Energy
Monitoring System with Open IoT Platform,” International Conference on Electronics,
Information, and Communication (ICEIC), Honolulu, HI, USA, 2018, pp. 1-2,
doi: 10.23919/ELINFOCOM.2018.8330550.
7. D. M. Kumar, R. Arthi, K. Dhanveer, V. P. Kumar, S. Mahidhar, “A Renewable Energy
Smart Metering System Using LoRa Network,” International Conference on Emerging
Systems and Intelligent Computing (ESIC), Bhubaneswar, India, 2024, pp. 626-630,
doi: 10.1109/ESIC60604.2024.10481548.

484
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

51. UV-C Based Plate Sterilizer

Suhas. R. Hatti, Abhishek C. Arahunasi
IEEE Student Member,
Dept. of Electronics and Communication Engineering
Basaveshwar Engineering College, Bagalkote, Karnataka India.
Vijaylakshmi S. J.
IEEE Professional Member,
Dept. of Electronics and Communication Engineering
Basaveshwar Engineering College, Bagalkote, Karnataka India.

ABSTRACT:

The aim of this paper is to reduce the transmission of sickness among the people through
the unclean plates used in large gatherings, marriage functions, temple fairs, etc. In large
gathering of people at fairs and marriages requires huge number of plates where cleaning
becomes the big task. This may lead to retain some microorganisms on plates which causes
infections with manual cleaning process. Due to the lack of cleaning, bacteria build up and
cause the spread of sickness. This proposed model works as both dish washer and sterilizing
unit which can be used in large gatherings.

KEYWORDS:

Dish washer, sterilizer, UV-C band, COVID, Infection.

I Introduction:

In today’s society, automation of everyday tasks has a large impact on people’s lives. India
is heavily populated country and has many traditional activities carried like, temple fairs,
marriages where food is served in large quantity. In such large gatherings the plates used
are just dipped in soap water to clean and served to the next people. This leads to retention
of microorganism on plates and further may cause infection. Due to the lack of proper
cleaning process, there is chance of spread of many viral diseases. After, COVID it is very
much necessary to maintain hygiene in such situation. Hence a low cost Ultra Violet (UV)
light based plate sterilizer is proposed in this paper.

The UV light C band based plate sterilizer is the ultimate solution for clean and sterilized
plates. This innovative home appliance is perfect for caterers, residences, and hostels alike.

485
International Journal of Research and Analysis in Science and Engineering

With a water spray, dish soap inlet, hot air drier, and sterilizer in one convenient unit, it
effectively removes any pathogens, ensuring hygienic and infection-free plates. Say
goodbye to worries about spreading infections and hello to a healthier dining experience.

Large gathering of people at fairs and marriages requires huge number of plates where
cleaning becomes the big task. This may lead to retain with some microorganisms on plates
which causes infections with manual cleaning process. Due to the lack of cleaning, bacteria
build up and cause the spread of sickness. This proposed model works as both dish washer
and sterilizing unit which can be used in large gatherings. Model consists of three main
units viz, washing unit, heating unit, drier and sterilizing unit.

A UV-C based Plate Sterilizer, utilizes UV-C sanitization to kill bacteria on any flat surface.
This process will eliminate any harsh chemicals required to sanitize and clean the plate
surface. This process will eliminate any harsh chemicals required to sanitize and clean a
surface.

II Literature Survey:

The literature survey covers the work done by researchers on the design and efficiency of
the dishwashing machines.

1. Hoak, D. Parker, D. Hermelink, A. American Council for an Energy Efficient Economy,

Washington, DC, August 2008. This Journal helps that present measurements of three
recent vintage dishwashers are very different efficiencies showing that while they
substantially more efficient than older dishwashers, those tested will still use electric
resistance elements for supplement heat, even when supplied by solar water heating
system producing very hot water.
2. Shilpa N Dehedkar- “Design of basic model of Semi-Automatic Dishwasher machine”.
(2016): This paper use brief idea and analysis of the Semiautomatic Dishwasher
machine. It also states the mechanisms incorporated in this model for process of
washing the dish. In this research the dishwashers operate with help of DC motor,
Universal motor, Conveyor belt and Microcontroller for time delay.
3. Shaila S. Hedaoo- “Design and Fabrication of Semi-Automatic Dish and Utensil
washing machine”. This paper discuss the main objective of Semi-Automatic
Dishwashing machine is to reduce the cost of fully automatic dishwashing machine and
giving good Cleaning Performance. It requires less energy and less water consumption.
Time of washing dish can be adjusted as per customer requirements.
4. PranaliKhatake- “Design of gears in semiautomatic dishwashing machine”. This paper
discusses about design of gear in semiautomatic dishwashing machine. The result
indicates that in India semiautomatic dishwashing machine are used than fully
automatic dishwashing machine as it is cheap, preferably gears are used in this
semiautomatic dishwashing machine with the belt drive for better life and high
efficiency.
5. Dhale A. D.- “Design and Development of semiautomatic dishwasher”. This paper
discusses about the design, construction and evaluation of dishwashing machine. The
capacity of machine was 20 plates per min (i.e. 1880 plates per hour). The design
dishwasher is very efficient and easy to operate.

486
 UV-C Based Plate Sterilizer

6. J. G. Gochran- “Dishwashing machine”. The paper gives brief idea describe about
improvement of dishwashing machine. It related to improvement in machine washing a
dishes in which continuous stream of either soap-soda or clean water is supply to crate
holding the rack or cage hot water is supplied to crate is rotate so as to bring the greater
portion thereof under water.
7. International Journal for Scientific Research and Development Vol.4, Issue 05,2016 |
ISSN (online): 2321- 0613. This journal explains that using Galvanized iron material
for inner and outer part, the overall weight of the assembly is also reduced. The capacity
of machine is to wash 24 pieces of dinner set at a time by using two rotary jet controlled
by single pump using parallel connection.
8. International Journal for Scientific Research and Development| Vol. 3, Issue 11,2016 |
ISSN (online):2321- 0613. This Journal represents the modified design of utensils
automatic washer machine. In this, the adjustable conveyor containing utensils tends to
rotate, and passing these utensils under three section scrubbing, water sprinkler and
cleaner. The dishwasher has made cleaning and drying dishes much easily and more
efficiently. Conveyor is rotated by using motors. This leads to making the design
simpler and better than the present dishwashers.
9. International Journal for Scientific Research and Development 2016 IJEDR | ISSN:
2321-9939.This journal suggests that this system multi jet technology is used to clean
Utensils. Any type of Utensils will be washed in our system; no electronic circuit will
be used. Multi jet system will be used to clean utensils from all side

III Problem Identification:

In India, washing of plates at large gatherings is done at least care Used plates are simply
dipped in soapy water and washed, which may cause pathogens or other bacteria to
accumulate on the plates. Following the effects of COVID, it is crucial to exercise caution
and keep oneself clean. An UV-C based plate sterilizer is proposed in this research work by
considering the following parameters.

1. Washing Time: The amount of time required for washing plates depends on factors
such as the type of dishwasher used, how dirty the plates are, and the efficiency of the
machine. On average, a standard dishwasher cycle takes approximately 1 to 2 hours to
complete. However, some industrial dishwashers can complete a cycle in 5 to 10
minutes.
2. Soap Consumption: The amount of soap needed to wash plates depends on the type of
soap or detergent used, the hardness of the water, and the dishwasher's settings.
Typically, a domestic dishwasher uses 3 to 6 liters of water per wash cycle and needs
roughly 15 to 30 grams of detergent. Industrial dishwashers may need larger amounts
of detergent depending on their capacity and the number of plates being washed.
3. Sterilization: Plates may not be totally sterilized even though dishwashers clean them
thoroughly. Typical operating temperatures for domestic dishwashers range from 113°F
to 167°F (45°C to 75°C), which is hot enough to effectively eliminate certain bacteria
as well as grease and grime. Nevertheless, additional techniques like high-temperature
rinsing or chemical sanitization might be necessary for total sterilization.

487
International Journal of Research and Analysis in Science and Engineering

4. Number of Plates: The amount of plates that can be cleaned in a single cycle depends
on the dishwasher's or sink's capacity that is being used for manual washing. 12 to 14
standard-sized dishes can normally fit in a cycle of a domestic dishwasher. In
commercial or industrial settings, depending on the size and kind of dishwasher, the
capacity can range from a few dozens to several hundred plates every cycle.

IV Proposed System:

The main concept of this proposed model is to reduce the transmission of sickness among
the people through the unclean plates used in large gatherings, marriage functions, temple
fairs, etc. For example, in large gathering of people at fairs and marriages requires huge
number of plates where cleaning becomes the big task. This may lead to retain with some
microorganisms on plates which causes infections with manual cleaning process. Due to the
lack of cleaning, bacteria build up and cause the spread of sickness. This proposed model
works as both dish washer and sterilizing unit which can be used in large gatherings.

Objectives:

• The primary objective of a UVC-based plate sterilizer is to inactivate or destroy

microorganisms, such as bacteria, viruses, and fungi, present on the surface of plates.
• The model objective is to effectively clean plates with less usage of water and
detergents.
• UVC-based plate sterilization in a dishwasher can save time and labor compared to
manual methods of plate sterilization.

A UVC-based plate sterilizer with a dishwasher is a system that is designed to clean and
sterilize plates using ultraviolet light

(UVC) and a dishwasher. The working of the proposed model is as shown in Figure 1:

Figure 1: Block Diagram

1. Pre-rinse: The plates would be pre-rinsed to remove any visible debris or food
particles.
2. Loading: The plates would then be loaded into the model.
3. Washing: The model would wash the plates using hot water and detergent, as is
typically done in a dishwasher.
4. Rinsing: After washing, the plates would be rinsed with clean water to remove any
remaining detergent.

488
 UV-C Based Plate Sterilizer

5. Drying: Before sterilization, the plates should be dried using hot air, as is typically done
in a dishwasher.
6. UVC Sterilization: Once the rinsing and drying process is complete, the plates would
be subjected to UVC light to sterilize them. UVC light has been shown to be effective
at killing a wide range of microorganisms, including bacteria and viruses. The UVC
light would be generated using special UVC lamps that are designed to emit a specific
wavelength of light that is effective at killing microorganisms.

The proposed system would be designed to be easy to use and efficient. It would be ideal
for use in commercial kitchens, restaurants, hospitals, and other environments where
hygiene is of the utmost importance. The UVC-based plate sterilizer with a dishwasher
would help to ensure that plates are thoroughly cleaned and sterilized, reducing the risk of
foodborne illnesses and infections.

V Flow Chart:

Figure 2

The UV-C based plate sterilizer's flow chart is shown in figure 2 above. First, plates will be
provided as the input to the washing machine, where any residual food on the plates will be
cleaned away using water.

Following that, plates are cleaned with clean water and soap. The sterilizing device receives
the clean plates, which are then sterilized by all germs on the surface of the plates. The use
of a hot air blower allows for the drying of plates before submitting them to the sterilizing
equipment. The clean, sterilized plates are received as an output from the sterilizing unit
uniqueness of the innovation:

489
International Journal of Research and Analysis in Science and Engineering

Figure 3: Internal Design of the Model

The above figure 3. shows the internal design of the proposed model. The UVC-based plate
sterilizer with a dishwasher is an advanced appliance that combines the functionality of a
dishwasher with the disinfecting power of UV-C (Ultraviolet-C) light technology. Using
UV-C technology, this sterilizer utilizes UV-C lamps that emit a specific wavelength of
light. This light effectively destroys bacteria, viruses, and other harmful microorganisms.
By integrating these UV-C lamps into the dishwasher, the sterilizer ensures that plates
receive an additional layer of disinfection during the cleaning process. Operating the UVC-
based plate sterilizer with a dishwasher is similar to using a regular dishwasher.

First step is to load dirty plates into the holder, then passed through spraying unit and when
the dishwasher cycle starts, the UV-C lamps are activated. The UV-C light penetrates the
surfaces of the plates, eliminating pathogens present on them. This proposed model is an
efficient and convenient solution for maintaining cleanliness and hygiene. Additionally, this
appliance is designed with energy efficiency and environmental considerations in mind.

Overall, the proposed UVC-based plate sterilizer with a dishwasher combines the cleaning
capabilities of a dishwasher with the disinfecting power of UV-C light technology. It offers
an effective and efficient solution for maintaining a high level of hygiene and cleanliness
by ensuring that the plates and dishes are sanitized and safe for use.

VI Application:

The UV-based plate sterilizer integrated with a dishwasher has the potential to find
applications in various industries and markets where hygiene, convenience, and efficiency
plays vital role. Some of the applications are as listed below:

1. Residential Kitchens: The residential kitchen market is a natural fit for this innovation,
as it caters to health-conscious families and individuals seeking convenient and
comprehensive solutions for dish cleaning and sterilization.

490
 UV-C Based Plate Sterilizer

2. Restaurants and Food Service: In the food service industry, where rapid turnaround
and stringent hygiene standards are crucial, this appliance could offer a highly efficient
and reliable means of cleaning and sterilizing dishes, utensils, and kitchen tools.
3. Healthcare Facilities: Hospitals, clinics, and other healthcare facilities require strict
sterilization protocols. The UV-based plate sterilizer with dishwasher could be used to
ensure that medical instruments and utensils are not only cleaned but also thoroughly
disinfected.
4. Catering and Event Planning: The catering and event planning industry often
involves handling large volumes of dishes. This appliance could streamline
dishwashing and sterilization processes for these businesses.
5. Educational Institutions: Schools, colleges, and universities could benefit from this
invention to maintain a higher level of hygiene in their cafeterias and dining halls.
6. Hospitality and Accommodation: Hotels, resorts, and cruise ships could integrate this
appliance to offer guests a heightened level of hygiene and cleanliness during their stay.

In each of these applications, the UV-based plate sterilizer with dishwasher could stand out
due to its unique combination of functions, providing enhanced hygiene, time savings, and
user convenience.

VII Conclusion:

In conclusion, UV-C based plate sterilizer is an effective and eco-friendly way of sterilizing
small objects such as plates, utensils, and other kitchen equipment. It uses ultraviolet light
to kill bacteria, virus, and other pathogens without the use of chemicals or harmful
emissions. The disinfection process is quick and efficient, taking only a few seconds to
complete, and it does not produce any harmful by-products. However, it is important to note
that UV-C light can be harmful to human skin and eyes, so precautions must be taken when
using the sterilizer. Overall, the UV-C based plate sterilizer provides a safe and efficient
way of sterilizing kitchen equipment and is a valuable addition to any household or
commercial kitchen.

References:

1. Hoak, D. Parker, D. Hermelink, A. American Council for an Energy Efficient Economy,

Washington, DC, August 2008.
2. Shilpa N. Dehedkar- “Design of basic model of semi-automatic dishwasher machine”.
National Conference On Innovative Trends in Science and Engineering (NC-ITSE 16)
Volume:4 Issue: 7 2016 pp.18-21
3. Shaila S. Hedaoo, Dr. C.C. Handa ,V. D. Dhopte- “Design and fabrication of
Semiautomatic dish and utensils washing machine. International Journal of Engineering
Development and Research Volume 4, Issue 3 2016 pp.292-296
4. Pranali Khatake- “Design of gears in Semiautomatic Dishwashing machine”.
International Journal for Scientific Research & development Vol.3, Issue 3, May-
June,2015 pp.108-112 987 | P a g e
5. Dhale A. D- “Design and development of semi automic”. International Journal of
engineering Development and Research (www.ijedr.org) 292 Volume 4, issue 3| ISSN;
2321-9939 IJEDR 1603047.

491
International Journal of Research and Analysis in Science and Engineering

6. J. G. Gochran- “Dishwashing Machine”. International Journal for Scientific Research

& Development Vol.3, issue 11,2016 | ISSN (online): 2321-0613
7. Pranali Khatake- “Design of gears in Semiautomatic Dishwashing machine”.
International Journal for Scientific Research & development Vol.3, Issue 3, May-
June,2015 pp.108-11
8. “Fabrication of semi-automatic Dishwasher and Utensils Washing Machine”.
International Journal of engineering Development and Research (www.ijedr.org) 292
Volume 4, issue 3| ISSN; 2321-9939 IJEDR 1603047
9. J. garis-Cochrane, “Dishwashing machine”, United states Patent Office, Patent no.
835,299, April 24 1917.
10. Chavan Shrikant et. al. “Automatic dishwashing Machine”. International Journal for
Scientific Research & Development Vol.3, issue 11,2016 | ISSN (online): 2321-0613,
[7] International Journal for Scientific Research and Development Vol.4, Issue 05,2016
| ISSN (online): 2321-0613
11. International Journal for Scientific Research and Development| Vol. 3, Issue 11,2016 |
ISSN (online):2321- 0613.
12. Darwish. S.M, Al-Samhan A.M, (2004). Peel and Shear Strength of Spot-Welded and
weld- bonded Dissimilar Thickness Joints, Process Technology 147, (pp. 51-59).
http://www.which.co.uk/reviews/dishwasherdetergent/page/featuresexplained/Lehigh-
Unilever study sheds light on mystery of cloudy wineglasses.
13. Rajput, R.K. (2003). A textbook of power plant engineering, (2nd ed.) New Delhi,
Lanmi Publications Ltd.
14. Spinks, A.T., Dunstan, R.H., Harrison, T., Coombes, P. & Kuczera, G. (2006). Thermal
inactivation of water-borne pathogenic and indicator bacteria at sub-boiling
temperatures. Water Res. 40
15. Stamminger, R., Badura, R., Broil, G., Doerr, S. & Elschenbroich, A. (2004) A
European comparison of cleaning dishes by hand [Online], University of Bonn:
PHPSESSID=6c9f9ccf3b8fa39071af22518408722e

492
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

52. Development of PD-PWM Technique for 5-

Level Cascaded H-Bridge Inverter using FPGA for
SPV Systems
Basavaraju S. Hadapad, Raghuram L. Naik,
Amrutha H. D., Asha Chougala, Kailash, Madhu D.
Department of Electrical and Electronics Engg.
Basaveshwar Engineering College, Bagalkote, India.

ABSTRACT:

In the recent years, multilevel inverters (MLI) are gaining importance in various industrial
applications and distributed energy sources due to its good waveform quality and low
switching stress for medium voltage to high power applications. Among different topologies
of MLIs, Cascaded H Bridge (CHB) is more suitable converter for SPV system as it provides
isolated DC input compared to other topologies. The conventional pulse width modulation
(PWM) control techniques have been used to control the CHB inverter. However, these
techniques are operated at higher switching frequency results in higher switching loss that
leads in harmonic content into the output voltage. In view of this, the paper presents Phase
Disposition Pulse Width Modulation (PD-PWM) technique to operate MLIs at fundamental
frequency with low harmonic content and less switching losses employed for SPV based
applications. The proposed model is implemented using Matlab/Simulink and FPGA Xilinx
tool box. Further, the model is validated on hardware test bench consisting of IGBT based
inverter connected through three phase uncontrolled rectifier and FPGA based controller.
It is observed from simulation and experimental results that the total harmonic distortion
content for voltage waveforms found to be 38.10% and 5.36 % respectively. Finally, it is
inferred that filtering requirement for 5-level is less as compared to that of two level or
three level inverters.

KEYWORDS:

Multilevel inverter, PWM technique, cascaded H bride inverter, SPV systems, FPGA
processor.

I. Introduction:

In recent years, people showing more interest in harnessing renewable energy sources over
the conventional sources of energy. Solar PV is one of the supreme gifted renewable energy

493
International Journal of Research and Analysis in Science and Engineering

sources. Owing to pollution and noise-free nature, solar PV is extensively adopted to satisfy
the increasing energy needs [1]. Generally, the solar PV systems are connected either to
utility grid or isolated systems. Grid-connected systems are highly preferred among them
because of their various advantages. Usually, an inverter is employed to transfer the power
from solar PV to the ac grid. Generally, a low-frequency transformer is used between the
inverter and ac grid while processing solar PV's power to the grid [2]. Various thyristor-
based grid tied inverters are reviewed in [3]. However, the inverters are preferred only to a
low power solar energy conversion system. An IoT enabled solar smart inverter is proposed
which is designed and carried out for 2-way communication between the environment and
end-user using Wi-Fi [4]. The inverter output is conveyed to the user. The end user having
the feasibility to manage the loads using mobile, allowing for more efficient energy use and
increased human comfort. Nevertheless, the smart system is applicable only for a stand-
alone system and it fails miserably for grid connected system. A novel solar inverter is
developed to mitigate the problems associated with conventional inverters such as high
switching losses, low efficiency and costly [5]. The designed converter is suited only for
distributed generation. MLIs have gained more attention in the field of high voltage and
medium power applications due to its advantages such as low voltage stress on
semiconductor devices, lower harmonic distortions, good electromagnetic compatibility and
reduced switching losses [6-10]. As the number of levels in inverter increases, the total
harmonic distortion in the output voltage reduces. The multilevel inverter has 3 topologies
such as diode clamped multilevel inverter, flying capacitors multilevel inverter, Cascaded
H-Bridge Multilevel Inverter. Among three topologies cascaded H Bridge Multilevel
inverter is more advantageous [11]. In Cascaded H bridge multilevel inverter, H-bridges are
connected in series and each H-bridge has separate DC source which can be obtained from
any natural sources like fuel cells or batteries to produce inverted ac output. Cascaded H
Bridge multilevel inverter is simpler than diode clamped and flying capacitor topologies
because of advantages such as automatic voltage sharing, smaller dv/dt stress, switching
redundancy, requirement of least number of components, no high rated capacitors and
diodes [12]. Cascaded H bridge multilevel inverter is suitable for battery, PV based
applications where isolated DC source is required [10].

The design of five level Cascaded H Bridge multilevel inverter using PD-PWM technique
for harmonic reduction is presented [13]. The THD content of the 5 level Cascaded H Bridge
Multilevel Inverter for different modulation index is compared by using mathematical
approach and MATLAB/SIMULINK. The performance of symmetrical and asymmetrical
single phase cascaded H bridge nine level inverter with respect to harmonic content, number
of switches and voltage stress across the switches with SPV panels is simulated using
MATLAB/SIMULINK [14]. It is observed from simulation result that, THD and voltage
stress across the switches is less in symmetrical type as compared to asymmetrical cascaded
H bridge nine level inverter. A detailed discussion on various MLI control schemes is
presented in [15]. A novel PWM technique to overcome the problems associated with
conventional PWM inverters is depicted in [16]. From aforementioned literatures, following
observations are made;

• Cascaded H-Bridge multilevel inverter is suitable for as it provides isolated DC source.

• Voltage stress and total harmonic distortion is less in symmetrical multilevel inverter
compared to asymmetrical multilevel inverter.

494
 Development of PD-PWM Technique for 5-Level Cascaded H-Bridge Inverter using FPGA…

• Phase Disposition Pulse width modulation technique is more convenient as it generates

less harmonic distortion in line to line voltage.
• Quality of the output voltage improves as the number of level increases.

In connection to this, a PD-PWM technique is implemented in this paper for 5-level

cascaded bridge MLIs employed for SPV based applications. This work has been performed
using FPGA processor as it offers more flexibility in designing of PWM generators for
power converters.

II. PWM techniques for 5-level CHB inverter:

The concept of MLI is based on connecting H bridge inverters in series to synthesize desired
voltage from separate DC source. A 5-level inverter can be developed to reduce switching
losses, THD and electromagnetic interference caused by the switching operation of the
power electronic devices. The circuit diagram of cascaded 5-level inverter is as shown in
Figure 1.

Figure 1: Single phase 5-level CHB inverter

Various PWM techniques are applied for CHB inverters. The most widely used PWM
techniques are Phase Disposition- PWM technique, Phase Opposition and Disposition
PWM technique (POD-PWM) and Alternate Phase Opposition and Disposition PWM
technique [16].

A. POD-PWM Technique:

In the POD-PWM pulse generation technique, n number of triangular carrier signals are
compared with a reference sine wave as shown in Figure 2. The generated output pulses
produce n-level output voltage waveform.

495
International Journal of Research and Analysis in Science and Engineering

Figure 2: Arrangement of Carrier Wave and Sine Wave

In this strategy, the triangular carrier signals above the zero reference are referred as upper
triangular and are represented as Tun = [Tu1, Tu2, Tu3]. The carrier signals below the zero
reference are referred as lower triangular and are represented as Tln = [Tl1, Tl2, Tl3]. Figure
3 illustrates the control logic schematic of the POD PWM scheme. If the magnitude of the
upper carrier is less than the reference sine wave, the comparator generates un and generates
(un – 1) if the condition is not satisfied. On the other hand, for the lower carrier being lower
than the reference signal, the comparator produces −(ln −1) and generates −ln if the
condition is not satisfied. The output waveforms have the same number of steps as of the
aggregated signal. The gate pulses are further acquired by decoding the aggregated signal
as per the switching sequence [17].

Figure 3: PWM signals generation by POD-PWM technique

B. Phase Opposite Disposition PWM Technique:

In this strategy, the triangular signal compares with sinusoidal reference signal to generate
the PWM signals [18]. In this method, multicarrier triangular signals are opposite to each
other as shown in Figure 4.

The PWM signals produces if sine wave is larger or equal to the carrier wave on the upper
side. Similarly, PWM generates if sine wave is lesser or equal to the carrier wave in the
lower side.
496
 Development of PD-PWM Technique for 5-Level Cascaded H-Bridge Inverter using FPGA…

Figure 4: Arrangement of carrier wave and sine wave

III. PD-PWM for 5-level CHB inverter:

PD-PWM is highly preferrable technique compare to others as it generates lowest harmonic

distortion in line-to-line voltage and hence PD-PWM strategy is presented in this paper. The
switches of cascaded H-bridge inverter operate as er the switching mechanism of PD-PWM
technique tabulated in Table-1 to achieve 5-level output voltage.

Table 3.1: Switching mechanism for 5 level CHB inverter as per PD-PWM technique

S1 S2 S3 S4 S5 S6 S7 S8 Output Voltage
1 1 0 0 1 1 0 0 + 2Vdc
1 1 0 0 0 1 0 1 +Vdc
0 1 0 1 0 1 0 1 0
0 0 1 1 0 1 0 1 - Vdc
0 0 1 1 0 0 1 1 - 2Vdc

PD-PWM signals for power electronic switches are generated by comparing the modulating
sinusoidal wave with the fixed amplitude reference signals. Further, generated pulses used
to control the state of switches used in the inverter. Figure 5 shows the controlling scheme
for gate pulse generation for H-bridge cells.

Figure 5: Control Scheme

497
International Journal of Research and Analysis in Science and Engineering

The generation of 5-level output voltage waveform requires comparison of sine wave with
the four fixed DC reference signals to obtain the pulses for all the switches of CHB inverter
as shown in the Figure 6.

Figure 6: Simulation model of PWM signals for 5-level voltage waveform

Simulation model of PD-PWM technique for 5-level cascaded H-bridge inverter is as shown
in Figure 7. It comprises of two isolated H-bridge inverters supplied by separate 20 W SPV
panels each are cascaded to generate 5-level voltage output. The output of inverter is
connected to 1 KVA R-L load.

Figure 7: Simulation model of PD-PWM signals for 5-level voltage waveform

IV. Results and Discussions:

SPV based 5-level CHB inverter consists of two independent solar panels having power
rating of 20 W each connected to the IGBT based two H bridge MLI separately. The
specifications of two solar panels used in this work are as shown in Table-2.

498
 Development of PD-PWM Technique for 5-Level Cascaded H-Bridge Inverter using FPGA…

Table-2: Specifications of SPV panel

Sl. No Particulars
1 Maximum Power (Pmax) 20 W
2 Current at Pmax (Imax) 1.11 A
3 Voltage at Pmax (Vmax) 18.10 V
4 Short-Circuit Current (Isc) 1.25 A
5 Open Circuit Voltage (Voc) 21.98 V

The simulation model of PD-PWM technique for the generation of 5-level output voltage
for CHB inverter is illustrated in Figure 7. The output voltage of the inverter measured
across a load of 1 KVA is as shown in Figure 8.

Figure 9 represents the harmonic spectra of load current and voltage of 5 level Cascaded H-
Bridge inverter respectively. Harmonics of the output is analyzed and simulated using FFT
block in simulink toolbox of Matlab. Output voltage of the proposed modulation technique
exhibits half wave odd symmetry. Hence only the positive half cycle need to be analyzed.

Figure 9: Harmonic spectra of voltage of 5 Level CHB

499
International Journal of Research and Analysis in Science and Engineering

According to the Fig.8 from the simulation result, the THD content for voltage observed is
38.10%. The simulation result is then validated by conducting an experiment using Xilinx
based FPGA processor. The experimental test bench is comprised of two isolated H bridge
inverters powered by two independent SPV panels of 20 W capacity. These inverters are
cascaded to achieve 5-level voltage across 1 KVA load. The switching signals by PD-PWM
technique is obtained by FPGA processor for the operation of IGBT switches in the inverter.
The generated switching signals are illustrated in Figure 10.

Figure 10: PWM signals generated by FPGA

The generated pulses are given to the gate terminals of IGBT switches for their operation in
the CHB inverter. The experimental test bench for conducting this experiment is installed
in the research laboratory of BEC, Bagalkot as shown in Figure 11.

Figure 11: Experimental test bench installed in BEC, Bagalkote

The experiment conducted when the solar radiation was measured to be 950 W/m2 at 12.30
pm on 20th May 2024. As the solar radiation was high, 5-level voltage waveform was
recorded across 1 KVA load as shown in Figure 12.
500
 Development of PD-PWM Technique for 5-Level Cascaded H-Bridge Inverter using FPGA…

Figure 12: Experimentally obtained 5-level voltage waveform

It is observed from the result that experimentally obtained voltage waveform is very close
to the simulated result. Smaller value of ripples was observed in experimentally obtained
voltage waveform owing to the variation in the solar radiation. However, ripple is too small
that do not affect the load performance. Harmonic analysis is also carried out for one cycle
of output voltage waveform achieved by PD-PWM technique. Output voltage of proposed
modulation technique exhibits half wave odd symmetry. Hence only positive half cycle was
analyzed. It is noticed that higher order harmonics are not severe as a lower, because of the
presence of inductance that causes higher order harmonics to damp out more quickly. THD
for the experimentally obtained waveform was found to be 5.36%.

Figure 13: THD of 5-level Voltage Waveform

V. Conclusion:

PD-PWM technique is designed and implemented for 5-level cascaded H-bridge inverter
employed for SPV based applications. The simulation of the proposed technique is carried
out using MATLAB/SIMULINK platform. In order to carry out the hardware
implementation, the entire algorithm is designed in xilinx tools using digital block sets. The
developed model is validated on hardware test bench consisting of IGBT based inverter for
1 KVA R-L load.
501
International Journal of Research and Analysis in Science and Engineering

From the FFT analysis it was observed that as the number of levels of CHB inverter
increases THD decreases. THD for 5 level CHB inverter is 38.10% from simulation and
THD content obtained from the experimental results is 5.36%. Both the results are compared
and validated. From the above results it is concluded that 5 level CHB is more suitable for
isolated DC inputs such SPV system.

References:

1. P. K. Chamarthi, A. Al-Durra, T. H. M. EL-Fouly and K. A. Jaafari, "A Novel Three-

Phase Transformerless Cascaded Multilevel Inverter Topology for Grid-Connected
Solar PV Applications," in IEEE Transactions on Industry Applications, vol. 57, no. 3,
pp. 2285-2297, May-June 2021, doi: 10.1109/TIA.2021.3057312.
2. K. Chen et al., "Cascaded iH6 multilevel inverter with leakage current reduction for
transformerless grid-connected photovoltaic system," 2017 IEEE 12th International
Conference on Power Electronics and Drive Systems (PEDS), Honolulu, HI, USA,
2017, pp. 613-617, doi: 10.1109/PEDS.2017.8289270.
3. M. R. Khalid, A. Sarwar and M. S. Jamil Asghar, "Review of Thyristor Based Grid Tied
Inverters for Solar PV Applications," 2019 IEEE 5th International Conference for
Convergence in Technology (I2CT), Bombay, India, 2019, pp. 1-5,
doi: 10.1109/I2CT45611.2019.9033746.
4. R. R. Rubia Gandhi, G. N. Danyashri, S. Aishvarya, S. D. Prabha, S. Arnesh and C.
Kathirvel, "PV based Smart Inverter for Reliable System," 2022 IEEE 2nd International
Conference on Mobile Networks and Wireless Communications (ICMNWC), Tumkur,
Karnataka, India, 2022, pp. 1-5, doi: 10.1109/ICMNWC56175.2022.10031717.
5. J. N. Mahajan and A. M. Jain, "Conversion of existing inverter into solar inverter," 2017
International Conference on Computing Methodologies and Communication (ICCMC),
Erode, India, 2017, pp. 859-862, doi: 10.1109/ICCMC.2017.8282587.
6. A. Mohan and P. Jayaprakash, "An Improved Seven Level Multilevel Inverter for Grid
Connected SPV Application," 2021 International Conference on Communication,
Control and Information Sciences (ICCISc), Idukki, India, 2021, pp. 1-6,
doi: 10.1109/ICCISc52257.2021.9485003.
7. C. Dinakaran and T. Padmavathi, "Simulation of Single-Phase Cascaded H-Bridge
Multilevel Inverter with Bidirectional Switches for Solar Photovoltaic Systems," 2022
International Conference on Breakthrough in Heuristics and Reciprocation of Advanced
Technologies (BHARAT), Visakhapatnam, India, 2022, pp. 25-30,
doi: 10.1109/BHARAT53139.2022.00017.
8. S. R. Aute and S. A. Naveed, "Simulation and Analysis of Multilevel Inverter Based
Solar PV System," 2019 3rd International Conference on Computing Methodologies
and Communication (ICCMC), Erode, India, 2019, pp. 557-559,
doi: 10.1109/ICCMC.2019.8819742.
9. A. K. Singh and R. Gupta, "Integrated Battery Management Configurations for
Standalone Solar PV fed CHBMLI," 2020 IEEE International Conference on Power
Electronics, Drives and Energy Systems (PEDES), Jaipur, India, 2020, pp. 1-5,
doi: 0.1109/PEDES49360.2020.9379876.
10. S. K. Baksi and R. K. Behera, "A Reduced Switch Count Seven-Level Boost ANPC
Based Grid Following Inverter Topology with Photovoltaic Integration," in IEEE
Transactions on Industry Applications, vol. 59, no. 4, pp. 4238-4251, July-Aug. 2023,
doi: 10.1109/TIA.2023.3259943.

502
 Development of PD-PWM Technique for 5-Level Cascaded H-Bridge Inverter using FPGA…

11. I. Sarkar and B. G. Fernandes, "A Hybrid Symmetric Cascaded H-Bridge Multilevel
Converter Topology," in IEEE Journal of Emerging and Selected Topics in Power
Electronics, vol. 11, no. 4, pp. 4032-4044, Aug. 2023,
doi: 10.1109/JESTPE.2021.3104222.
12. F. Esmaeili, A. Saadat and H. R. Koofigar, "A High-Gain Multilevel Inverter Topology
with Low Harmonics for PV Grid-Tied Applications," 2024 28th International
Electrical Power Distribution Conference (EPDC), Zanjan, Iran, Islamic Republic of,
2024, pp. 1-6, doi: 10.1109/EPDC62178.2024.10571704.
13. Swathy, S., Niveditha, N., Chandragupta Mauryan, K.S. (2020). Design of Five-Level
Cascaded H-Bridge Multilevel Inverter. In: Saini, H., Srinivas, T., Vinod Kumar, D.,
Chandragupta Mauryan, K. (eds) Innovations in Electrical and Electronics Engineering.
Lecture Notes in Electrical Engineering, vol 626. Springer, Singapore.
https://doi.org/10.1007/978-981-15-2256-7_7.
14. N. Booma and N. Sridhar, "Nine level cascaded H-bridge multilevel DC-link inverter,"
2011 International Conference on Emerging Trends in Electrical and Computer
Technology, Nagercoil, India, 2011, pp. 315-320,
doi: 10.1109/ICETECT.2011.5760135.
15. E. Chaharmahali, M. Sepehrinour, H. Karimi and A. Siadatan, "Harmonic Distortion
Optimization of Multilevel H-bridge Inverter with Delta-Connected," 2023 13th Smart
Grid Conference (SGC), Tehran, Iran, Islamic Republic of, 2023, pp. 1-5, doi:
10.1109/SGC61621.2023.10459277.
16. A. Ishtiaq, M. Asim and M. Anis, "A Brief Review on Multi level Inverter Methodology
Topologies and Techniques," 2022 2nd International Conference on Emerging Frontiers
in Electrical and Electronic Technologies (ICEFEET), Patna, India, 2022, pp. 1-5, doi:
10.1109/ICEFEET51821.2022.9848098.
17. R. Sarker, "Phase Disposition PWM (PD-PWM) Technique to Minimize WTHD from
a Three-Phase NPC Multilevel Voltage Source Inverter," 2020 IEEE 1st International
Conference for Convergence in Engineering (ICCE), Kolkata, India, 2020, pp. 220-224,
doi: 10.1109/ICCE50343.2020.9290697.
18. Sourabh Rathore, Mukesh Kumar Kirar and S. K Bhardwaj, “Simulation of Cascaded
H- Bridge Multilevel Inverter using PD, POD, APOD Techniques”, Electrical &
Computer Engineering: An International Journal (ECIJ) Volume 4, Number 3,
September 2015.

503
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

53. A Review of Various Attacks and Detection

Methods in Internet of Medical Things (IoMT)
Systems
Benazir Muntasher
Department of Computer Science and Engineering SECAB Institute of
Engineering and Technology Vijayapur, India.
Mahabaleshwar S. Kakkasageri
Department of Electronics and Communication Engineering
Basaveshwar Engineering College Bagalkote, India
Engineering and Technology Vijayapur, India.

ABSTRACT:

The Internet of Medical Things (IoMT) revolutionizes healthcare by connecting medical

devices and enabling seamless data exchange. However, this connectivity introduces
significant security risks. This review paper discusses various attacks targeting IoMT
systems and the corresponding detection and correction methods leveraging Machine
Learning (ML), Deep Learning (DL), Artificial Intelligence (AI), and Blockchain
technologies. Recent advancements from 2022 to 2024 are emphasized, highlighting their
contributions to enhancing the security and resilience of IoMT systems.

I Introduction:

The Internet of Medical Things (IoMT) offers transformative potential in healthcare by

enabling real-time monitoring, efficient data management, and improved patient outcomes.
However, its interconnected nature also introduces significant security challenges. This
review has highlighted various cyberattacks targeting IoMT systems, including malware,
phishing, Denial of Service (DoS), Man-in-the-Middle (MitM), data breaches, device
hijacking, side-channel attacks, supply chain attacks, physical attacks, and exploitation of
vulnerabilities. To mitigate these threats, advanced detection methods leveraging Machine
Learning (ML), Deep Learning (DL), Artificial Intelligence (AI), and Blockchain
technologies have been developed. ML algorithms, such as Random Forest and Support
Vector Machines, along with DL models like Convolutional Neural Networks (CNN) and
Recurrent Neural Networks (RNN), provide robust capabilities for detecting anomalies and
suspicious patterns.

504
 A Review of Various Attacks and Detection Methods in Internet of Medical Things (IoMT) Systems

AI techniques, including heuristic analysis and Natural Language Processing (NLP),

enhance the detection of complex threats, while Blockchain ensures data integrity and
security through immutable logs and secure identity verification.

Effective correction methods are crucial for maintaining the security of IoMT systems.
These include regular patch management, antivirus software, network segmentation, user
training, multi-factor authentication (MFA), traffic filtering, load balancing, rate limiting,
strong encryption, access control, regular audits, firmware updates, secure supply chain
practices, and physical access controls.

The ongoing development and implementation of these advanced technologies are essential
to address emerging threats and ensure the resilience of IoMT in the evolving cyber
landscape. As the IoMT ecosystem continues to expand, a comprehensive approach to
security, integrating advanced detection and correction methods, will be vital to
safeguarding patient data and maintaining the integrity of healthcare system.

II Attacks and Detection Mechanisms in IoMT:

The Internet of Medical Things (IoMT) refers to the interconnected ecosystem of medical
devices and applications that communicate health data over networks. IoMT facilitates real-
time monitoring, improves patient outcomes, and enhances the efficiency of healthcare
delivery. However, the increased connectivity also introduces vulnerabilities, making IoMT
systems prime targets for cyber-attacks. Various Attacks on IoMT are as follows.

Malware Attacks: Malware infiltrates IoMT devices, compromising their functionality and
data integrity. Common malware types include viruses, worms, trojans, and ransomware
[1].

• Phishing Attacks: Phishing attacks deceive users into providing sensitive information
through fraudulent emails or websites, leading to unauthorized access to patient data
and systems [2].
• Denial of Service (DoS) Attacks: DoS attacks overwhelm IoMT networks with
excessive traffic, rendering devices and services unavailable to legitimate users [3].
• Man-in-the-Middle (MitM) Attacks: MitM attacks intercept and alter communication
between IoMT devices and servers, compromising data integrity and confidentiality [4].
• Data Breaches: Data breaches occur when unauthorized individuals gain access to
sensitive health information due to weak security protocols [5].
• Device Hijacking: Attackers take control of IoMT devices, potentially disrupting
services or launching further attacks [6].
• Side-Channel Attacks: Side-channel attacks exploit physical characteristics of IoMT
devices, such as power consumption or electromagnetic emissions, to extract sensitive
information [7].
• Supply Chain Attacks: Compromises within the supply chain can introduce
vulnerabilities into IoMT devices before deployment [8].
• Physical Attacks: Physical attacks involve direct access to IoMT devices, leading to
tampering and unauthorized data access [9].

505
International Journal of Research and Analysis in Science and Engineering

• Exploitation of Vulnerabilities: Unpatched software and hardware vulnerabilities can

be exploited to gain unauthorized access or control [10].
• Detection Methods: Machine Learning (ML): ML algorithms like Random Forest,
SVM, and K-Means Clustering are used to detect anomalies and suspicious patterns in
IoMT data [11].
• Deep Learning (DL): DL models such as CNN, RNN, and Auto encoders provide
advanced pattern recognition capabilities for detecting sophisticated attacks [12].
• Artificial Intelligence (AI): AI techniques, including heuristic analysis and Natural
Language Processing (NLP), enhance the detection of complex threats [13].
• Blockchain: Blockchain technology ensures data integrity and security through
immutable logs, secures identity verification, and decentralized filtering [14].

Correction Methods: Possible correction methods for these attacks are as follows.

• Patch Management: Regular updates to software and firmware address vulnerabilities

and improve security [15].
• Antivirus Software: Antivirus software protects IoMT devices from known malware
threats [16].
• Network Segmentation: Network segmentation isolates different segments of the
network to contain and mitigate attacks [17].
• User Training: Educating users on security best practices helps prevent phishing and
social engineering attacks [18].
• Multi-Factor Authentication (MFA): MFA adds layers of security for accessing
IoMT systems [19].
• Traffic Filtering: Traffic filtering blocks malicious traffic to prevent DoS attacks [20].
• Load Balancing: Load balancing distributes network load to mitigate the impact of
DoS attacks [21].
• Rate Limiting: Rate limiting controls the rate of requests to prevent system overload
[22].
• Strong Encryption: Strong encryption ensures secure communication channels and
data transmission [23].
• Access Control: Access control restricts access to authorized users only, ensuring data
security and privacy.
• Regular Audits: Regular security audits maintain compliance and identify potential
vulnerabilities.
• Firm ware Updates: Firmware updates keep device software up-to-date to protect
against known vulnerabilities.
• Secure Supply Chain Practices: Secure supply chain practices ensure the integrity and
security of IoMT devices throughout the supply chain.
• Physical Access Controls: Physical access controls prevent unauthorized physical
access to devices.

506
A Review of Various Attacks and Detection Methods in Internet of Medical Things (IoMT) Systems

Table I classifies various attacks and corresponding machine learning (ML), deep
learning (DL) and artificial intelligence (AI) detection methods in IoMT:

Attack Type Detection Method Type Description Examples References

Malware Signature-Based Rule- Uses Antivirus, IDS [35][36][15]
Attacks Detection Based predefined
signatures to
detect known
malware.
Anomaly Detection ML Detects Random Forests, [5][6][1]
deviations from SVMs
normal
behavior
indicating
malware.
Behavioral Analysis ML Monitors and User behavior [31][36][15]
analyzes device analytics
behavior for
signs of
malware
activity.
Phishing Heuristic Analysis Rule- Uses heuristic Heuristic antivirus [25][16][10]
Attacks Based rules to identify
phishing
attempts.
Natural Language AI Analyzes text Spam filters, Email [24][26][18]
Processing (NLP) in emails and classifiers
messages for
phishing
indicators.
Denial of Traffic Analysis ML Monitors Neural Networks, [35][36][15]
Service (DoS) network traffic SVMs
to detect
abnormal
patterns
indicating DoS
attacks.
Anomaly Detection DL Uses deep Auto encoders, [35][36][15]
learning models LSTM networks
to identify
unusual traffic
patterns.
Man-in-the- Anomaly Detection ML Detects Random Forests, [35][36][15]
Middle (MitM) anomalies in Isolation Forests
communication
patterns
indicating
MitM attacks.
Encryption and AI Uses AI to Blockchain-based [30][36][15]
Authentication enhance systems
507
International Journal of Research and Analysis in Science and Engineering

Attack Type Detection Method Type Description Examples References

encryption and
authentication
mechanisms.
Data Breaches Behavioral AI Uses biometric User behavior [10][31][36]
Biometrics patterns to analytics
detect
unauthorized
access.
Log Analysis ML Analyzes SIEM systems [36][31][35]
system logs to
identify
unusual access
patterns.
Device Anomaly Detection ML Monitors SVMs, Random [1][6][15]
Hijacking device behavior Forests
for signs of
hijacking.
Behavioral Analysis ML Analyzes Device behavior [9][19][35]
device usage analytics
patterns to
detect hijacking
attempts.
Side-Channel Anomaly Detection ML Detects Neural Networks, [10][12][23]
Attacks abnormal side- SVMs
channel signals.
Supply Chain Blockchain AI Ensures data Secure data [31][36][27]
Attacks Technology integrity and management
traceability in
the supply
chain.
Physical Anomaly Detection ML Monitors Random Forests, [3][24][29]
Attacks physical access Isolation Forests
and tampering
attempts.
Exploitation of Anomaly Detection ML Identifies Neural Networks, [35][36][15]
Vulnerabilities exploitation Auto encoders
attempts
through
unusual system
behavior.
Brute Force Anomaly Detection ML Detects Random Forests, [35][36][31]
Attacks repeated failed SVMs
access
attempts.
Eavesdropping Anomaly Detection ML Monitors Neural Networks, [5][6][15]
network Isolation Forests
communication
for unusual
patterns.

508
 A Review of Various Attacks and Detection Methods in Internet of Medical Things (IoMT) Systems

Attack Type Detection Method Type Description Examples References

Replay Attacks Anomaly Detection ML Detects SVMs, Random [33][36][35]
repeated data Forests
transmission
patterns.
Firmware and Anomaly Detection ML Monitors Neural Networks, [17][3][5]
Software software Auto encoders
Attacks behavior for
signs of
malicious
activity.

Table I: Various attacks and corresponding machine learning (ML), deep learning
(DL) and artificial intelligence (AI) detection methods in IoMT

The IoMT ecosystem faces diverse cyber security threats that can compromise patient safety
and data integrity. Advanced detection methods leveraging ML, DL, AI, and blockchain
play a crucial role in identifying and mitigating these threats.

Effective correction methods further enhance the security of IoMT systems. Ongoing
research and development are essential to address emerging threats and ensure the resilience
of IoMT in the evolving cyber landscape. Table II presents summary of attacks its detection
and correction mechanisms.

Table 2: Summary of Attacks, its detection and correction

Attack Type Detection Methods Correction Methods

Malware Attacks ML: Random Forest, SVM DL: CNN, Patch Management,
RNNAI: Heuristic Analysis Antivirus Software,
Blockchain: Immutable Logs Network Segmentation
Phishing Attacks ML: Logistic Regression, Naïve Bayes User Training, Multi-
DL: LSTM Factor Authentication
(MFA)
AI: NLP for Phishing Detection
Blockchain: Secure Identity
Verification
Denial of Service ML: K-Means Clustering, Decision Traffic Filtering, Load
(DoS) Trees DL: Auto encoders AI: Traffic Balancing, Rate Limiting
Analysis Blockchain: Decentralized
Filtering
Man-in-the- ML: Anomaly Detection Models DL: Strong Encryption, Secure
Middle (MitM) Gated Recurrent Units (GRU AI: Communication Protocols,
Encryption Protocol Analysis Certificate Pinning
Blockchain: Secure Communication
Protocols

509
International Journal of Research and Analysis in Science and Engineering

Attack Type Detection Methods Correction Methods

Data Breaches ML: Anomaly Detection, Isolation Data Encryption, Access
Forest DL: Variation Auto encoders Control, Regular Security
(VAE), AI: Blockchain for Data Audits
Integrity
Device Hijacking ML: Support Vector Machines (SVM), Device Authentication,
DL: Deep Belief Networks (DBN), AI: Firmware Updates,
Federated Learning, Blockchain: Network Segmentation
Device Authentication
Side-Channel ML: PCA for Anomaly Detection, DL: Hardware Shields,
Attacks Recurrent Neural Networks (RNN), AI: Electromagnetic Isolation
Behavioral Analysis, Blockchain: Power Management
Secure Data Transmission Strategies
Supply Chain ML: Anomaly Detection, K-Means Secure Supply Chain
Attacks Clustering, DL: Convolutional Neural Practices, Regular Audits,
Networks (CNN), AI: Blockchain for Firmware Verification
Supply Chain Integrity
Physical Attacks ML: Random Forests, DL: LSTM Physical Access Controls,
Networks, AI: Physical Security Tamper-Evident Seals,
Analysis, Blockchain: Tamper-Evident Surveillance Systems
Logs
Exploitation of ML: Anomaly Detection, Decision Regular Software Updates,
Vulnerabilities Trees, DL: Auto encoders, AI: Code Reviews, Penetration
Heuristic Analysis, Blockchain: Secure Testing
Code Updates

III Conclusion:

The security of IoMT systems is crucial for maintaining patient safety and the integrity of
medical data. A multi-layered approach, incorporating effective detection methods and
robust correction strategies, is essential to protect against various types of attacks. As the
IoMT landscape continues to evolve, ongoing vigilance and adaptation to emerging threats
will be key to safeguarding these vital technologies.

References:

1. Ghubaish, A., Salman, T., Zolanvari, M., Unal, D., Al-Ali, A., & Jain, R. (2021). Recent
Advances in the Internet-of-Medical-Things (IoMT) Systems Security. IEEE Internet
of Things Journal, 8(11), 8707–8718. doi:10.1109/jiot.2020.3045653
2. Thomasian, Nicole M., and Eli Y. Adashi. "Cyber security in the internet of medical
things." Health Policy and Technology 10.3 (2021):100549.
3. Chen, Chong, et al. "Avoid attacks: A Federated Data Sanitization Defense in IoMT
Systems." IEEE INFOCOM 2023-IEEE Conference on Computer Communications
Workshops (INFOCOM WKSHPS). IEEE,2023.

510
 A Review of Various Attacks and Detection Methods in Internet of Medical Things (IoMT) Systems

4. Peddle, Brandon & Lu, Wei & Yu, Qiaoyan. (2023). Detecting DDoS Attacks in the
Internet of Medical Things Through Machine Learning-Based Classification.
10.1007/978-3-031-47126-1_13.
5. 2020 Vision: A Review of Major IT & Cyber Security Issues Affecting Healthcare.
Cyber MDX, New York, NY, USA. Accessed: Nov. 18, 2020. [Online]. Available:
https://www.cybermdx.com/resources/2020-visionreview-major-healthcare-it-
cybersec-issues
6. W. Maddox. Why Medical Data Is 50 Times More Valuable Than a Credit Card.
Accessed: Nov. 18, 2020. [Online]. Available: https://www.dmagazine.com/healthcare-
business/2019/10/why-medicaldata-is-50-times-more-valuable-than-a-credit-card/
7. Information Assurance, United States Naval Acad., Annapolis, MD, USA. Accessed:
Nov. 18, 2020. [Online]. Available:
https://www.usna.edu/Users/cs/wcbrown/courses/si110AY13S/lec/l21/lec.html
8. J.-P. A. Yaacoub et al., “Securing Internet of Medical Things systems: Limitations,
issues and recommendations,” Future Gener. Comput. Syst., vol. 105, pp. 581–606,
Apr. 2020.
9. A. Vyas and S. Pal, “Preventing security and privacy attacks in WBAN,” in Handbook
of Computer Networks and Cyber Security, B. Gupta, G. Perez, D. Agrawal, and D.
Gupta, Eds. Cham, Switzerland: Springer, 2020, pp. 201–225.
10. D. Bhushan and R. Agrawal, “Security challenges for designing wearable and IoT
solutions,” in A Handbook of Internet of Things in Biomedical and Cyber Physical
System, (Intelligent Systems Reference Library), vol. 165, V. Balas, V. Solanki, R.
Kumar, and M. Ahad, Eds. Cham, Switzerland: Springer, 2020, pp. 109–138.
11. F. Pesapane, M. B. Suter, M. Codari, F. Patella, C. Volonté, and F. Sardanelli, “Chapter
52—Regulatory issues for artificial intelligence in radiology,” in Precision Medicine
for Investigators, Practitioners and Providers, J. Faintuch and S. Faintuch, Eds. London,
U.K.: Academic,2020, pp. 533–543
12. J. Sengupta, S. Ruj, and S. D. Bit, “A comprehensive survey on attacks, security issues
and blockchain solutions for IoT and IIoT,” J. Net w. Comput. Appl., vol. 149, pp. 1–
20, Jan. 2020.
13. S. M. Tahsien, H. Karimipour, and P. Spachos, “Machine learning based solutions for
security of Internet of Things (IoT): A survey,” J. Net w. Comput. Appl., vol. 161, pp.
1–18, Jul. 2020.
14. L. Xiao, Y. Li, G. Han, G. Liu, and W. Zhuang, “PHY-layers proofing detection with
reinforcement learning in wireless networks,” IEEE Trans. Veh. Technol., vol.65,
no.12, pp.10037–10047, Dec. 2016.
15. Implant (Medicine), Wikipedia. Accessed: Nov.18, 2020. [Online]. Available:
https://en.wikipedia.org/wiki/Implant_(medicine)
16. J. E. Ferguson and A. D. Redish, “Wireless communication with implanted medical
devices using the conductive properties of the body,” Expert Rev. Med. Devices, vol.
8, no. 4, pp. 427–433, 2011.
17. Pacemaker, Mayo Clinic, Rochester, MN, USA. Accessed: Nov. 18, 2020. [Online].
Available:
https://www.mayoclinic.org/testsprocedures/pacemaker/about/pac-20384689
18. A. Phaneuf. Latest Trends in Medical Monitoring Devices and Wearable Health
Technology. Accessed: Nov. 18, 2020. [Online]. Available:
https://www.businessinsider.com/wearable-technologyhealthcare-medical-devices

511
International Journal of Research and Analysis in Science and Engineering

19. Heart Health Notifications on Your Apple Watch, Apple, Cupertino, CA, USA.
Accessed: Nov. 18, 2020. [Online]. Available: https://support.apple.com/en-
us/HT208931
20. A. Kos, V. Milutinovic, and A. Umek, “Challenges in wireless communication for
connected sensors and wearable devices used in sport bio feedback applications,”
Future Gener. Comput. Syst., vol. 92, pp.582–592, Mar.2019.
21. H. Jahankhani and J. Ibarra, “Digital forensic investigation for the Internet of Medical
Things (IoMT),” J. Forensic Legal Invest. Sci., vol.5, p. 029, Aug. 2019.
22. Pacing Systems— zure|Medtronic, Medtronic, Dublin, Ireland. Accessed: Nov. 18,
2020. [Online]. Available: https://europe.medtronic.com/xd-en/healthcare-
professionals/products/cardiacrhythm/pacemakers/azure.html
23. Medical Device Radio Communications Service (MedRadio), Federal Commun.
Commission (FCC), Washington, DC, USA. Accessed: Nov. 18, 2020. [Online].
Available:https://www.fcc.gov/medical-deviceradiocommunications-service-medradio
24. P. Kasyoka, M. Kimwele, and S. Mbandu Angolo, “Certificate less pairing-free
authentication scheme for wireless body area network in healthcare management
system,” J. Med. Eng. Technol., vol. 44, no. 1, pp. 12–19, 2020.
25. T. Belkhouja, S. Sorour, and M. S. Hefeida, “Role-based hierarchical medical data
encryption for implantable medical devices,” in Proc. IEEE Global Commun. Conf.
(GLOBECOM), Waikoloa, HI, USA, Dec. 2019, pp. 1–6.
26. Chinese Remainder Theorem, Wikipedia. Accessed: Nov. 18, 2020. [Online].
Available: https://en.wikipedia.org/wiki/Chinese_remainder_theorem
27. Received Signal Strength Indication, Wikipedia. Accessed: Nov.18, 2020. [Online].
Available: https://en.wikipedia.org/wiki/Received_signal_strength_indication
28. M. F. Awan, K. Kansanen, S. Perez-Simbor, C. Garcia-Pardo, S. Castelló-Palacios, and
N. Cardona, “RSS-based secret key generation in wireless in-body networks,” in Proc.
13th Int. Symp. Med. Inf. Commun. Technol. (ISMICT), Oslo, Norway, 2019, pp. 1–6.
29. Cryptographic Hash Function, Wikipedia. Accessed: Nov. 18,2020. [Online].
Available: https://en.wikipedia.org/wiki/Cryptographic_hash_ function
30. XOR Gate, Wikipedia. Accessed: Nov. 18, 2020. [Online]. Available:
https://en.wikipedia.org/wiki/XOR_gate
31. B. A. Alzahrani, A. Irshad, A. Albeshri, and K. Alsubhi, “A provably secure and
lightweight patient-healthcare authentication protocol in wireless body area networks,”
Wireless Pers. Commun., pp.1–23, Mar. 2020.
32. Z. Xu, C. Xu, W. Liang, J. Xu, and H. Chen, “A lightweight mutual authentication and
key agreement scheme for medical Internet of Things,” IEEE Access, vol. 7, pp. 53922–
53931, 2019.
33. Y. Sun and B. Lo, “An artificial neural network framework for gait-based biometrics,”
IEEE J. Biomed. Health Inform., vol. 23, no. 3, pp. 987–998, May 2019.
34. V. H. Tutari, B. Das, and D. R. Chowdhury, “A continuous role-based authentication
scheme and data transmission protocol for implantable medical devices,” in Proc. 2nd
Int. Conf. Adv. Comput. Commun. Paradigms (ICACCP), Gangtok, India, Feb. 2019,
pp. 1–6.
35. S. Maji et al., “A low-power dual-factor authentication unit for secure implantable
devices,” in Proc. IEEE Custom Integr. Circuits Conf. (CICC), Boston, MA, USA,
2020, pp. 1–4.
36. RSA (Cryptosystem), Wikipedia. Accessed: Nov. 18, 2020. [Online]. Available:
https://en.wikipedia.org/wiki/RSA_(cryptosystem)

512
 A Review of Various Attacks and Detection Methods in Internet of Medical Things (IoMT) Systems

37. Elliptic-Curve Cryptography, Wikipedia. Accessed: Nov. 18, 2020. [Online].

Available: https://en.wikipedia.org/wiki/Ellipticcurve_cryptography
38. V. J. Jariwala and D. C. Jinwala, “Chapter 4—Adaptable SDA: Secure data aggregation
framework in wireless body area networks,” in Wearable and Implantable Medical
Devices, vol. 7, N. Dey, A. S. Ashour, S. James Fong, and C. Bhatt, Eds. London, U.K.:
Academic,2020, pp. 79–114.
39. T. Bhatia, A. K. Verma, and G. Sharma, “Towards a secure incremental proxy re-
encryption for e-healthcare data sharing in mobile cloud computing,” Concurrency
Comput. Pract. Exp., vol. 32, no. 5, pp.1–16,2020.
40. Zettabyte, Wikipedia. Accessed: Nov. 18,2020. [Online]. Available:
https://en.wikipedia.org/wiki/Zettabyte
41. Homomorphic Encryption, Wikipedia. Accessed: Nov. 18, 2020. [Online]. Available:
https://en.wikipedia.org/wiki/Homomorphic_encryption
42. G. Kalyani and S. Chaudhari, “An efficient approach for enhancing security in Internet
of Things using the optimum authentication key,” Int. J. Comput. Appl., vol. 42, no. 3,
pp. 306–314, 2020.
43. Digital Signature, Wikipedia. Accessed: Nov. 18, 2020. [Online]. Available:
https://en.wikipedia.org/wiki/Digital_signature
44. C. Easttom and N. Mei, “Mitigating implanted medical device cyber security risks,” in
Proc. IEEE 10th Annu. Ubiquitous Comput. Electron. Mobile Commun. Conf.
(UEMCON), New York, NY, USA,2019, pp. 0145–0148.
45. A. Kumari, S. Jangirala, M. Y. Abbasi, V. Kumar, and M. Alam, “ESEAP: ECC based
secure and efficient mutual authentication protocol using smart card,” J. Inf. Security
Appl., vol. 51, pp. 1–12, Apr. 2020.
46. G. Zheng et al., “Finger-to-heart(F2H): Authentication for wireless implantable medical
devices,” IEEE J. Biomed. Health Inform., vol. 23, no. 4, pp. 1546–1557, Jul. 2019.
47. G. Zheng, W. Yang, M. Johnstone, R. Shankaran, and C. Valli, “3— Securing the
elderly in cyberspace with fingerprints,” in Assistive Technology for the Elderly, N. K.
Suryadevara and S. C. Mukhopadhyay, Eds. London, U.K.: Academic, 2020, pp. 59–
79.
48. N. Ibtiheland S. M. Hadj, “Smart ECG monitoring through IoT,” in Smart Medical Data
Sensing and IoT Systems Design in Healthcare, C. Chin May Ed. Hershey, PA, USA:
IGI Global, 2020, pp. 224–246.
49. IoT and Cloud Tools to Build Your Business, Ubidots, Medellín, Colombia. Accessed:
Nov. 18, 2020. [Online].
50. W. Youssef, A. O. Zaid, M. S. Mourali, and M. H. Kammoun, “RFID based system for
secure logistic management of implantable medical devices in tunisian health centers,”
in Proc. IEEE Int. Smart Cities Conf. (ISC2), Casablanca, Morocco, 2019, pp. 83–86.

513
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

54. Hybrid SegAN Fuzzy–Unet and DKN

Classification for Crop Field Change Detection
using Satellite Images
Ms. Mulik Madhuri Balasaheb
Research Scholar VTU Belgavi, BEC Research Center
Electronics and Computer Engineering,
Sharad Institute of Technology College of Engineering,
Yadrav – Ichalkaranji, India.
Dr. Kulkarni P. N.
Department of Electronics & Communication Engg.
Basaveshwar Engineering College Bagalkot, India.
Dr. V. Jayshree
Electrical Engineering & PG Coordinator,
Electronics Engineering, DKTE Society’s Textile and
Engineering Institute Ichalkaranji, India.
Abstract:

Changes in land-cover detection are important landscape characteristics that affect

ecosystem conditions and function. Cropland detection is commonly used to find the
agricultural area with the help of satellite images. The proposed approach detects the crop
field using temporal change detection and satellite images. The satellite image-based crop
field change detection is performed using the proposed hybrid DL-based segmentation, and
deep Kronecker network (DKN) based crop field change detection.

Keywords:

Deep Kronecker Network, crop, satellite images, segmentation, U-net.

Introduction:

Remotely sensed satellite imagery has been widely used in agriculture, geology, forestry,
regional planning, and many other fields to analyse and manage natural resources and
human activities. The advancement of imaging technologies has resulted in satellite
deployment in recent years with very high spatial resolution imaging systems, enabling
satellite images to measure small things on the surface up to 0.5 m and give more precise
earth observation [1].
514
 Hybrid SegAN Fuzzy–Unet and DKN Classification for Crop Field Change Detection…

Quantitative information was frequently extracted from satellite photos to monitor the
Earth's environment on both a spatial and temporal scale [2]. Because of remote sensing can
repeatedly and reliably scan the land surface across wide areas, it has been utilized over the
past forty years to monitor land cover and changes [3]. Decision-makers have more and
more access to satellite image processing for mapping vegetation and planning future
expansion and development. More than ten years ago, remote sensing was discovered as a
technique for performance evaluation [4]. For many practical applications, hyperspectral
imaging (HSI) analysis is essential.

High spectral resolution images produced by HSI offer a greater amount of information.
Numerous applications, including object segmentation and land cover categorization, have
made use of this large amount of data. Many techniques for HSI classification rely on the
use of manually created features, which lessens the undesirable impacts of pixels and
enhances the accuracy of HSI classification [5]. A high-resolution satellite imaging system's
architecture is developed through a series of trade-offs between various performance
specifications and design trade-offs, such as speed of data transfer, picture compression,
revisit time, and off-nadir viewing angle [6].

The availability of high-resolution Satellite Image Time Series (SITS) is a result of

technological advancements in Earth observation instruments, as well as better temporal
and geographical resolutions. [7]. Global change detection has been done by Land use and
land-cover change detection [8]. However, due to trade, socioeconomic shocks, institutional
frameworks, and land-use policies, agricultural land abandonment is also a regular land-use
change process in many regions of the world [9]. Before and after the change are crucial for
segmentation-based change detection, and the temporal information present in the full-time
series has not been adequately utilized. Incorporating both spatial and temporal
segmentation is advantageous as it aids in comprehending the dynamics of land-use
systems, including agricultural abandonment [9].

Land cover detection reflects possible natural and social mechanisms in agricultural
development. With the advent of several earth observation satellites, image quality is
constantly being improved concerning spatial, spectral, and temporal resolutions [10].
Using remotely sensed data, land-use change analysis has been used at diverse scales and
geographic resolutions throughout the world's ecosystems and nations. In addition, ten-year
research of the dynamics of crop area, production, and pricing in the study area included
farm-level surveys, regional crop areas, productivity evaluation using remote sensing
techniques, and national statistics [11].

Deep Learning (DL) has gained popularity in the past few years for picture comprehension
problems, such as remote sensing image comprehension. Supervised satellite image analysis
has been conducted using DL-based change detection techniques. When the labelled multi-
temporal training data are available, a supervised deep learning approach is selected for CD.
Additionally, the architecture of a recurrent convolutional neural network (ReCNN) is used
to extract joint spectral-spatial-temporal information [12]. Convolutional neural networks
(CNNs) were used in several remote sensing applications, including HSI analysis and the
classification of high-resolution images, which sparked interest in deep learning in the field
of remote sensing [13].

515
International Journal of Research and Analysis in Science and Engineering

I. Scientific Rationale / Relevance:

The rise of crop production depends mainly on the agricultural land cover. Any area's
economic development must be measured and assessed, and identifying shifts in agriculture
throughout time is crucial. Therefore, a lot of techniques were developed for crop field
change detection. Still, various obstacles were faced such as lengthy processing time,
uncertainty of satellite sensors, higher error rate, and so on. Meshkini, K., et al. [12], used
Three-Dimensional (3D) CNN. The reliability of the model was high, and it offered a lower
value of false alarms. This model could not forecast the other areas with various change
classes by using the fine-tuning strategy used for 3D CNN feature extraction. Park, S., et al.
[2] implemented High spatial resolution image fusion using object-based weighting
(HIFOW).

To restore the detailed spatial patterns within crop fields with less spectral distortion and
clear crop boundaries the HIFOW model was employed. However, HIFOW did not specify
how to fix the multi-sensor images' radiometric discrepancy. Xi, W., et al. [3] developed a
Spatiotemporal Cube (ST-Cubes) based Spatiotemporal Contextual method. The dense
satellite image time series (SITS) data for the intra-annual land cover mapping was given
by this model.

Thus, it allowed for more precise temporal scale land cover dynamics investigations. This
approach manually determined the “trial-and-error” values. Therefore, the time
consumption was high. Liu, W., et al [9] used Attention-based multiscale transformer
network (AMTNet). It provided high-resolution optical remote sensing data analysis despite
complicated textures, varying seasons, and shifting climates.

The AMTNet was not applicable for weakly supervised learning for change detection tasks.
Maddala, V.K.S., et al. [14] implemented Multisensory Data and Cross-Validation
Technique. To confirm the accuracy of the observation, this technique was employed to get
feedback from several farms. It was devoid of the extra components that would have
improved the overall administration of a wide range of agricultural operations. Tang, C., et
al. [15] used the Hybrid Dilated HDC-Siam model was utilized for resolving the issues with
pseudo-change and intra-class change in mining regions' change detection tasks.

This model did not consider the hybrid optimization approach to train the HDC-Siam to
attain more accurate performance. Bhattacharjee, S., et al. [16] implemented Remote
sensing (RS) and geographic information system (GIS) techniques. In this model, the
gradual change of deep water into shallow water over time was enabled by the local
community, and it expanded agricultural lands and activities during the dry season.

This approach failed to examine and mitigate the potential impacts of altering the
hydrological feature. RF was created by Zhang, T., et al. [17] using Bayesian optimization
of parameters. With reduced noise and incorrectly categorized pixels, this approach
produced the most effective land cover visualization results. This model's shortcoming was
that it evaluated the algorithms' performance using a limited number of training datasets.
Furthermore, this model's performance in real-time was subpar.

516
 Hybrid SegAN Fuzzy–Unet and DKN Classification for Crop Field Change Detection…

Moreover, the following challenges are faced by the existing approaches:

• Diverse bandwidths along with spectral responses are observed in [2]'s multi-sensor
images for identical spectral bands. The spatiotemporal image fusion (STIF) prediction
ability was impacted by these various radiometric features of multi-sensor pictures.
• While the parameters in the temporal interaction potential's transition matrix were
discovered empirically, the spatial interaction potential of ST-Cubes [3] was obtained
using a trial-and-error method. To detect crop fields in other places with distinct change
types, these criteria might not be suitable.
• The AMTNet model [9] provided channel attention for enhancing the feature
representation of changed areas, while the transformer module provided the long-range
dependencies with ease manner. Still, it failed to combine the CNNs, attention
mechanisms, and transformers to attain even more precise estimation.
• The RF classifier with Bayesian hyper parameter optimization provided land cover
classification performance in a fast manner. However, this model did not consider the
hybrid Metaheuristic optimization for getting better performance.

The spatial and temporal features of crop growth status and production were revealed by
the satellite-driven crop-field detection, which yielded cropland statistics at local, regional,
and global scales. However, there has been a shift in the climate, and there were insufficient
quantitative and reliable ways to guarantee the accuracy of crop data, which limited the
application of crop field observation and had undetermined and unfavourable effects.

II. Significant Contribution Including Innovation:

In the proposed crop field change detection model, the satellite image collected from the
dataset will be preprocessed using the Kaman filter, and the preprocessed image will be
subjected to the proposed hybrid fuzzy SegAN fuzzy–Unet based segmentation process.
Here, the SegAN Fuzzy–Unet will be created by the merging of SegAN, and Unet, where
the layers will be modified by the fuzzy concept.

III. Methodology:

To comprehend and identify ecosystems, resources, and environmental processes, one must
have a detailed grasp of land cover, which is provided by satellite images land covers
provide essential information for understanding and detecting ecosystems, resources, and
environmental dynamics, in which the satellite image provides a precise view of land cover.

Moreover, crop field detection is essential for smart agriculture. Therefore, this work
develops the temporal changes in crop fields using satellite images. Initially, the satellite
image at time 1 will be subjected to the image pre-processing stage, where the noisy as well
as redundant data will be removed using the Kalman [18] filter. Afterward, the segmentation
will be performed using the proposed hybrid SegAN. Moreover, the SegAN Fuzzy–Unet
will be designed by the combination of SegAN [19] and Unet [20], where layers will be
modified using the fuzzy concept. Therefore, the segmented image 1 will be obtained.

517
International Journal of Research and Analysis in Science and Engineering

Figure 1: Block diagram of proposed hybrid SegAN Fuzzy–Unet-based crop field

change detection model

Moreover, a similar process will be performed for the satellite images with time 2, in which
the segmented image 2 will be obtained. Following that, the segmented images of 1 and 2
will be applied to the segment mapping process with the aid of the Deep Kronecker network
(DKN) [21]. Finally, crop field change detection will be obtained. Moreover, the proposed
model will be implemented in the PYTHON tool using the BHOONIDHI@NRSC/ISRO
dataset [22]. In addition, metrics like error, accuracy, and Jaccard coefficient will be utilized
to compute the model's performance. Figure 1 discusses the suggested model's block
diagram.

IV Result:

This section demonstrates the experimental results of the crop change detection using a
hybrid SegAN Fuzzy–Unet-based model using the images obtained from the dataset,
BHOONIDHI@NRSC/ISRO dataset [22] Maps. Figure 2 shows the sample results of the
experiment.

The proposed hybrid SegAN Fuzzy–Unet-based crop field change detection model offered
good performance on agricultural fields. The crop field change detection is performed using
the BHOONIDHI@NRSC/ISRO dataset. In the existing schemes, the major issue is the
flawless results with better accuracy. However, the proposed method provides an accuracy
of more than 98%. The proposed approach considerably reduced the computational
complexity and thus offered enhanced accuracy in prediction. The hybrid DL network
approach delivers an effective segmentation process of a crop field.

518
 Hybrid SegAN Fuzzy–Unet and DKN Classification for Crop Field Change Detection…

a) b) c)

d) e) f)

Figure 2: Sample results of the experiment using - a) Original image of reference image
b) Original image test image, c) crop change reported in reference image d) crop
change reported in test change, e) Segmented result using a reference image, f)
Segmented result using a test image

Table I

Methods Accuracy Sensitivity Specificity

STIF 0.948 0.954 0.764
ST-Cubes 0.936 0.962 0.624
AMTNet model 0.924 0.976 0.841
SegAN Fuzzy–Unet 0.981 0.985 0.887

V Conclusion:

Variations in the detection of land cover are significant features of the landscape that impact
the state and functionality of ecosystems. Using satellite imagery, the agricultural region is
frequently located using the process of "cropland detection." The suggested method uses
satellite imagery and temporal change detection to identify the crop field. The suggested
hybrid DL-based segmentation and deep Kronecker network (DKN) based crop field change
detection are used to carry out the satellite image-based crop field change detection.

519
International Journal of Research and Analysis in Science and Engineering

References:

1. Xia, G.S., Yang, W., Delon, J., Gousseau, Y., Sun, H. and Maître, H., “Structural high-
resolution satellite image indexing”, 2009.
2. Park, S., Park, N.W. and Na, S.I., “An Object-Based Weighting Approach to
Spatiotemporal Fusion of High Spatial Resolution Satellite Images for Small-Scale
Cropland Monitoring”, Agronomy, vol.12, no.10, pp.2572, 2022.
3. Xi, W., Du, S., Du, S., Zhang, X. and Gu, H., “Intra-annual land cover mapping and
dynamics analysis with dense satellite image time series: A spatiotemporal cube based
spatiotemporal contextual method”, GIScience & Remote Sensing, vol.58, no.7,
pp.1195-1218, 2021.
4. Shwetank, J.K. and Bhatia, K., “Review of rice crop identification and classification
using hyper-spectral image processing system”, International Journal of Computer
Science & Communication, vol.1, no.1, pp.253-258, 2010.
5. Iyer, P., Sriram, A. and Lal, S., “Deep learning ensemble method for classification of
satellite hyperspectral images”, Remote Sensing Applications: Society and
Environment, vol.23, pp.100580, 2021.
6. Dial, G., Bowen, H., Gerlach, F., Grodecki, J. and Oleszczuk, R., “IKONOS satellite,
imagery, and products”, Remote sensing of Environment, vol.88, no.1-2, pp.23-36,
2003.
7. Solano-Correa, Y.T., Bovolo, F., Bruzzone, L. and Fernández-Prieto, D., “Spatio-
temporal evolution of crop fields in Sentinel-2 Satellite Image Time Series”, In
Proceedings of the 2017 9th International Workshop on the Analysis of Multitemporal
Remote Sensing Images (MultiTemp), pp. 1-4, 2017.
8. Aguiar, S., Texeira, M., Garibaldi, L.A. and Jobbágy, E.G., “Global changes in crop
diversity: trade rather than production enriches supply”, Global Food Security, vol.26,
pp.100385, 2020.
9. Yin, H., Prishchepov, A.V., Kuemmerle, T., Bleyhl, B., Buchner, J. and Radeloff, V.C.,
“Mapping agricultural land abandonment from spatial and temporal segmentation of
Landsat time series”, Remote sensing of environment, vol.210, pp.12-24, 2018.
10. Bhattacharjee, S., Islam, M.T., Kabir, M.E. and Kabir, M.M., “Land-use and land-cover
change detection in a north-eastern wetland ecosystem of Bangladesh using remote
sensing and GIS techniques”, Earth Systems and Environment, vol.5, no.2, pp.319-340,
2021.
11. Gumma, M.K., Gauchan, D., Nelson, A., Pandey, S. and Rala, A., “Temporal changes
in the rice-growing area and their impact on livelihood over a decade: A case study of
Nepal”, Agriculture, Ecosystems & Environment, vol.142, pp.3-4, pp.382-392, 2011.
12. Meshkini, K., Bovolo, F. and Bruzzone, L., “A 3D CNN approach for change detection
in HR satellite image time series based on a pre-trained 2D CNN”, The International
Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences,
vol.43, pp.143-150, 2022.
13. Rawat, A., Kumar, A., Upadhyay, P. and Kumar, S., “Deep learning-based models for
temporal satellite data processing: Classification of paddy transplanted fields”,
Ecological Informatics, vol.61, pp.101214, 2021.
14. Maddala, V.K.S., Jayarajan, K., Braveen, M., Walia, R., Krishna, P., Ponnusamy, S.
and Kaliyaperumal, K., “Multisensor data and cross-validation technique for merging
temporal images for the agricultural performance monitoring system”, Journal of Food
Quality, 2022.

520
 Hybrid SegAN Fuzzy–Unet and DKN Classification for Crop Field Change Detection…

15. Tang, C., Zhang, Z., He, G., Long, T., Wang, G., Wei, M. and She, W., “An improved
fully convolution network model for change detection in mining areas using sentinel-2
images”, Remote Sensing Letters, vol.12, no.7, pp.684-694, 2021.
16. Bhattacharjee, S., Islam, M.T., Kabir, M.E. and Kabir, M.M., “Land-use and land-cover
change detection in a north-eastern wetland ecosystem of Bangladesh using remote
sensing and GIS techniques”, Earth Systems and Environment, vol.5, no.2, pp.319-340,
2021.
17. Zhang, T., Su, J., Xu, Z., Luo, Y. and Li, J., “Sentinel-2 satellite imagery for urban land
cover classification by optimized random forest classifier”, Applied Sciences, vol.11,
no.2, pp.543, 2021.
18. Pakrashi, A. and Chaudhuri, B.B., “A Kalman filtering induced heuristic optimization
based partitional data clustering”, Information Sciences, vol.369, pp.704-717, 2016.
19. Xue, Y., Xu, T., Zhang, H., Long, L.R. and Huang, X., “Segan: Adversarial network
with multi-scale l 1 loss for medical image segmentation”, Neuroinformatics, vol.16,
pp.383-392, 2018.
20. Jiao, L., Huo, L., Hu, C., and Tang, P., “Refined UNet: UNet-based refinement network
for cloud and shadow precise segmentation”, Remote Sensing, vol.12, no.12, pp.2001,
2020.
21. Feng, L. and Yang, G., “Deep Kronecker Network”, arXiv preprint arXiv:2210.13327,
2022.
22. The BHOONIDHI@NRSC/ISRO dataset taken from ":
https://bhoonidhi.nrsc.gov.in/bhoonidhi/index.html", accessed on October 2023.

521
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

55. IOT Based Pollution Monitoring and

Controlling System
Sarojini B. K., Suhas Hatti
Department Electronics and Communication Engineering
Basaveshwar Engineering College, Bagalkote, India.
Abstract:

Industrial pollution has become a major concern worldwide, and effective monitoring and
controlling systems are needed to prevent environmental damage and protect human health.
In this proposed system, the designing of an IoT based system for monitoring and
controlling industrial pollution is carried out. The system consists of a network of sensors,
a data acquisition unit, a cloud-based platform, and a user interface. The sensors are
deployed at different locations in the industrial area to measure various pollutions such as
air, water, and noise. The data acquisition unit collects the sensor data and sends it to the
cloud-based platform for storage and analysis. The controlling unit is responsible for
activating various controlling mechanisms that is intimation to Pollution Control Board
(PCB) which is responsible to remove the power supply to the industry that is polluting
based on the pollution levels. PCB re-enables the power only if a penalty is paid by polluting
industry and once the power is re-enabled, the process/work of industry continues. The user
interface that is Blynk IOT app provides a dashboard that displays the real-time data. The
dashboard also allows users to configure the system, set threshold values, and generate
reports. It can help industries to comply with environmental regulations, reduce pollution
levels, and improve their sustainability.

Keywords:

Pollution, sensor, IoT, monitor, control.

I. Introduction:

The pollution, whether in the form of air, water and sound can have detrimental effects on
human health. Monitoring pollution levels helps to identify areas with high concentrations
of pollutants, enabling authorities that is Pollution Control Board (PCB) to take appropriate
actions to reduce the emission of pollutants and thereby reduce the pollution [5]. By
controlling pollution sources, such as industrial emissions or hazardous waste disposal, the
risk of adverse health effects can be reduced.

522
 IOT Based Pollution Monitoring and Controlling System

Pollution poses a significant threat to the natural environment, including ecosystems,

wildlife, and biodiversity. Monitoring pollution allows the identification of sensitive areas
and ecosystems at risk. Controlling pollution helps to prevent ecological imbalances, habitat
destruction, and the decline of species populations.

Industrial pollution is a major environmental challenge that affects the health of human
being, animals, and ecosystems. To tackle this problem, many countries have established
regulations and guidelines for monitoring and controlling industrial pollution. However, the
implementation and enforcement of these regulations would be difficult without proper
monitoring and control systems in place.

The proposed system aimed at developing a monitoring and control system for industrial
pollution. This can help industries comply with regulations and reduce their impact on the
environment. It involves sensors for monitoring and measuring air, water, and noise
pollution. This system also includes implementation of control measures to eliminate
pollution at the source.

The controlling unit is responsible for activating various controlling mechanisms that is
intimation to Pollution Control Board (PCB) which is responsible to remove the power
supply to the industry that is polluting based on the pollution levels. PCB re-enables the
power only if a penalty is paid by polluting industry and once the power is re-enabled the
process /work of industry continues.

Overall, a proposed system is focused on monitoring and controlling of three major

pollutions i.e. air, water and noise that can have significant environmental, economic, and
social advantages. It can help to protect human health, reduce the impact on ecosystems,
and promote sustainable industrial practices.

II. Related Work:

K. Rambabu, et al in the year 2016 [1] proposed a system "Industrial pollution monitoring
using Lab-View". Their system monitors pollution due to carbon monoxide and aimed at
only one parameter i.e. air pollution monitoring and did not take the measures for control
action.

Zumyla, et al in the year 2019 [2] proposed “IoT based Industrial Pollution Monitoring
System a smart city pollution monitoring and control system using IoT and big data
analytics. The system includes various sensors to monitor air pollutant. The collected data
is transmitted to a cloud-based platform for further analysis and visualization. The system
also includes an alerting mechanism to notify authorities of high pollution levels. This
proposed system design aimed at only one parameter that is air pollution monitoring and
did not take the measures for control action.

Demetillo, et al. in the year 2019 [3] proposed a system a system for monitoring water
quality in a large aquatic area using WSN technology. This system measures dissolved
oxygen temperature and pH of water. This system did not take the measures for control
action also did not consider any control actions to reduce the pollution.

523
International Journal of Research and Analysis in Science and Engineering

Increasing pollution day by day in atmospheric condition in the form of air water and noise
leads to increase in global warming, destruction of Ozone layer and also creating a lot of
health issues for the human being.

There are several steps taken to control the problem but the methods are not so effective
because some designs included either of parameters, air, water or noise pollution monitoring
and controlling and some included only monitoring and use huge amount of human
interaction increasing the negligence which leads to the problem to be not solved in a
systematic way.

The Pollution Control Board (PCB) department is not so active to visit each industry
manually and to take the action of the particular industry who exceeds these predefined safe
values of pollution limits. Hence this proposed system encompasses all three parameters air,
water and noise pollution monitoring and controlling.

III. Proposed System:

The block diagram of proposed pollution monitoring and controlling system is shown in
figure 1. The proposed system consists of Node MCU, CO2 (MQ135) sensor module, water
pollution detector (IR photo diode sensor), noise pollution detector (microphone sound
sensor). These sensors are used to measure the levels of pollutants such as carbon monoxide,
carbon dioxide, nitrogen dioxide, sulphur dioxide, and particulate matter in the air, water
pollution and noise level.

The sensors are connected to a microcontroller that collects the data and transmits it to the
controlling unit using a communication module such as Wi-Fi or Bluetooth. The controlling
unit receives the data and processes it using a software program to determine the pollution
levels.

Figure 1: Blockdiagram of Pollution Monitoring and Controlling system

A. Node MCU:

Node MCU as shown in Figure 2 and 3 is an open-source development board specially used
for IoT based Applications. It has firmware that runs on the ESP8266 Wi-Fi SoC, and
hardware that is based on the ESP-12 module.

524
 IOT Based Pollution Monitoring and Controlling System

Figure 2: Node MCU

Figure 3: Node MCU (labelled)

B. MQ135 Sensor:

The MQ135 sensor shown in Figure 4 is based on metal oxide semiconductor (MOS)
technology, where the sensing element is made of a thin film of tin dioxide (SnO2) that is
heated to a high temperature. When the sensor encounters a gas, the gas molecules are
absorbed by the sensing element, causing a resistance change in the sensor (the change in
resistance is then detected and converted into an electrical signal by the sensor) [4, 9, 15].
These changes are measured by the sensor's circuitry, which can provide an output voltage
proportional to concentration of gas being detected.

The MQ-135 sensor is sensitive to a wide range of gases such as ammonia, carbon dioxide,
nitrogen oxides, and benzene. It has a detection range of 10 to 1000 ppm (parts per million)
for most gases, with a maximum sensitivity to carbon dioxide gas. The sensor has a response
time of less than 10 seconds and a recovery time of around 30 seconds.

525
International Journal of Research and Analysis in Science and Engineering

Figure 4: Air Pollution Sensor

C. IR Photo Diode Sensor:

An IR (infrared) photo diode sensor used as water pollution sensor shown in Figure 5 is a
type of electronic component that is used to detect infrared light. The sensor comprises of a
photo diode, which is a semiconductor device that generates a small electrical current when
it is exposed to light [7, 8]. Infrared light has a longer wavelength than visible light and is
not visible to the human eye. IR photo diode sensors are commonly used in these devices to
detect the infrared signals and convert them into electrical signals that can be processed by
the device. This Sensor module works on the principle of Reflection of Infrared Rays from
the incident surface. A continuous beam of IR rays is emitted by the IR LED. Whenever a
reflecting surface comes in front of the Receiver (photo diode), these rays are reflected and
captured. Based on the amount of light that is reflected, amount of water polluted is decided.
When the pollution level in industry exceeds the predefined safe value output goes high.
This signal is connected to IOT NODE MCU unit to take further actions.

Figure 5: IR Photo diode sensor

D. Microphone Sound Sensor:

A microphone sound sensor used as noise pollution sensor shown in Figure 6 is a device
that is designed to detect and measure sound waves in the environment. The sensor contains
a small diaphragm that vibrates in response to sound waves, which is then converted into
an electrical signal that can be processed by a processor [9, 12]. When noise inside the
industry is detected by using the microphone, the sound signals are converted into electrical
signals, because of this condition the potential difference between two inputs at comparator
also changes and the comparator output goes from its low to high state. This signal is
coupled to IOT Node MCU unit to take further actions.
526
 IOT Based Pollution Monitoring and Controlling System

Figure 6: Noise Pollution Sensor

IV Results and Discussion:

The IOT based pollution monitoring and controlling system has yielded promising results
in addressing air, water, and noise pollution. Through the implementation of various
strategies and measures, significant progress has been noticed in reducing pollution levels
and controlling its adverse impacts by removing the power supply to the industry which is
polluting.

A. Air Pollution:

For detection of air pollution MQ135 sensor is used, which measures the concentration of
gases in the atmosphere like benzene, ammonia, oxides of sulphur, oxides of nitrogen, and
oxides of carbon, the major pollutant contributing for air pollution is CO2 and it is detected
by this sensor. The output is analog voltage which is converted to ppm and is displayed in
cloud interfaced platform used in this system that is Blynk app (Blynk IOT ESP8266). The
concentration of CO2 level is displayed in Gauge as shown in Figure 7, the detection range
of MQ135 sensor is 10-1000 ppm, if the value of concentration of CO2 exceeds 400 ppm
(400ppm-threshold value), it is considered as polluted air and control measure of
enable/disable of power supply control action is taken by pollution control board (PCB).
This control action results in a noticeable decrease in pollutant concentrations in the
environment. Stricter emission standards, and enforcement of regulations will contribute to
improve air quality, reducing the risks of respiratory and cardiovascular diseases.

Figure 7: Air Pollution level indicator

The sensitivity range of MQ135 sensor is 10-1000ppm and the threshold value is 400ppm.
If the gas concentration value exceeds 400ppm then it is viewed as air pollution.
527
International Journal of Research and Analysis in Science and Engineering

B. Water Pollution:

For detection of water pollution “IR photo diode sensor” is used, which indicates the water
quality. The detection of water quality is dependent on intensity of light received by the
photo diode when IR LED transmits the light, this sensor is placed in the waste water outlet
of the industry. The output of sensor is digital, if value is digital ‘1’ then it is the indication
of polluted water, for digital value ‘0’ gives the indication of pure water as shown in Figure
8. This result will be displayed in the Blynk app (Blynk IOT ESP8266) in Gauge. Based on
these results the Pollution Control Board (PCB) is responsible for taking the control action,
that is to enable/disable the power supply. The Strict regulations on industrial effluents will
lead to a significant improvement in water quality. Water bodies which will contaminate
can be reduced, ensuring a healthier aquatic ecosystem and a safer water supply for
communities.

Figure 8: Water Pollution Level Indicator

C. Noise Pollution:

For the detection of noise, Microphone sound sensor is used, when sound in the air hits the
diaphragm the plates of sensor vibrates and the distance between two plates varies, this
gives the sound level in air. The sensitivity range of sensor is 48- 52 dB. The output of
sensor is digital. If the output of sensor is digital ‘1’ then it is the indication of noise
pollution, digital value ‘0’ indicates there is no noise pollution as indicated in Figure 9. This
result will be displayed in the Blynk app (Blynk IOT ESP8266) in Gauge. Based on these
results the Pollution Control Board (PCB) is responsible for taking the control action that is
to enable/disable the power supply.

Figure 9: Noise Pollution Level Indicator

528
 IOT Based Pollution Monitoring and Controlling System

Measures to control noise pollution yields positive outcomes, leading to a quieter and more
peaceful environment. Implementation of noise reduction measures, such as sound barriers
and zoning regulations will contribute to improvement in quality of life for the people living
near noisy areas.

Based on the above results, the Pollution Control Board (PCB) cut off the power supply to
the polluting industry if the output of the sensor exceeds the values set by the PCB and the
power supply is reconnected if the penalty is paid to the PCB.

The results of the pollution monitoring and controlling demonstrate the effectiveness of
comprehensive strategies in protecting the environment. While significant progress has been
made, continuous efforts, and ongoing monitoring are essential to sustain the achieved
results and further improve environmental quality. Model of IOT based Pollution
Monitoring and Controlling System is shown in Figure 10.

Figure 10: Model of IOT Based Pollution Monitoring and Controlling System

V. Conclusions:

The pollution monitoring and controlling has been a significant endeavor in addressing the
growing concerns regarding pollution. The proposed system aimed at monitoring various
forms of pollution such as air, water and noise and implement effective control measures to
avoid the harmful effects on the environment and human health. This proposed system
emphasizes the need for sustained monitoring and adaptation to ensure long term success in
pollution monitoring and controlling. Regular monitoring and evaluation of pollution levels
will help to identify emerging issues and allow for the implementation of timely
interventions and adjustments of control measures. The control measure taken is to cut-off
the power supply by Pollution Control board (PCB) for the industries which are polluting
and re-enable the power supply if and only if the polluting industry has paid the penalty.

References:

1. K. Rambabu, B. Vasu, M. Raju, T. Vinod, Dr. M. C. Chinaaiah, “Industrial Pollution

Monitoring System Using LabVIEW”, International Journal of Scientific Research in
Science, Engineering and Technology IJSRSET, May 2016.
529
International Journal of Research and Analysis in Science and Engineering

2. Zumyla Shanaz F1, Prem Kumar S R, Rahul R, Rajesh Kumar M, Santhosh Kumar C
“IoT based Industrial Pollution Monitoring System” International Research Journal of
Engineering and Technology (IRJET) Vol 06 Issue 03 |Mar 2019, pp 2028-2041
3. Alexander T. Demetillo, Michelle V. Japitana, and Evelyn B. Taboada “IoT based
Industrial Pollution Monitoring System” Sustainable Environment Research, 2019 pp
1-9
4. Octavian Postolache, “Smart Sensors Network for Air Quality Monitoring
Applications”, IEEE Transactions on Instrumentation and Measurement, September
2009.
5. S. Muthukumar, “IOT Based Air Pollution Monitoring and Controlling System”,
Conference: International conference on Inventive Research in Computing
Applications(ICIRCA), July 2018.
6. H M Shamitha, Eramma, “IOT Controlled Advanced Industrial Pollution Monitoring
and Controlling System”, International Research Journal of Engineering and
Technology, October 2021.
7. Alexander T. D, Michelle V. J, Evelyn B. Taboda, “A System for monitoring water
quality in a large aquatic area using wireless sensor network technology”, Demetillo et
al. Sustainable Environment Research, December 2019.
8. S. Geetha, S. Gouthami, “Internet of things enabled real time water quality monitoring
system”, Article, July 2017.
9. Shraddha, Pooja, A D Sonawan, “IoT Based Air and Noise Pollution Monitoring
System”, Article in International Journal for Modern Trends in Science and
Technology, July 2021.
10. V Kalyan Kumar, Dr. Jeevan K M, Arun Gowda K, Harsha Vardhan, “IOT based
Industrial Pollution Monitoring System”.
11. Nicolas Maisonneuve, Matthias Stevens “Citizen Noise Pollution Monitoring”,
Proceedings of the 10th Annual International conference on Digital Government
Research, Partnerships for public Innovation, Dg. O 2009, Puebla, Mexico, May 2009.
12. Shreya Srivastava and Tanuj Sharma, “IoT-Based Industrial Pollution Monitoring and
Controlling System using Machine Learning", Published in IEEE Transactions on
Industrial Informatics, 2021.
13. Xiaoli Zhang, Xiaohui Li, and Li Li, "Industrial Pollution Monitoring and Control
System Based on IoT and Cloud Computing", published in Sensors, 2020.
14. D. Sivakumar and K. S. Deepak, "A Framework for Industrial Pollution Monitoring and
Control Using IoT and Cloud Computing", published an International Journal of
Innovative Technology and Exploring Engineering, 2019.
15. N. N. Thakur and P. R. Deshmukh, "IoT-Based Real-Time Air Pollution Monitoring
and Controlling System for Industrial Area", published in International Journal of
Advanced Research in Computer Science and Software Engineering, 2018.
16. R. Anitha and K. Prabhu,"IoT-Based Industrial Pollution Monitoring and Control
System", published in International Journal of Pure and Applied Mathematics, 2017.

530
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

56. Design and Implementation of Multi Prime

Mover Coupled Novel Water Pump for Small Scale
Irrigation Requirements
Vijaykumar Purutageri, Tejashwini Bagewadi,
Basanagouda Ronad
Department of E&EE, Basaveshwar Engineering College,
Bagalkote, Karnataka, India.
Mukta Bannur
Department of E&EE, BLDEA’s CET
Vijayapura, Karnataka, India.
Abstract:

The deficient in electricity and high diesel costs affects the pumping requirements of
irrigation systems for small scale agriculture. Therefore, using solar energy and other
sustainable water pumping methods is a promising alternative to conventional electricity
and diesel based pumping systems. This paper presents a hand held & pedal driven novel
type water pump for irrigation system, which can also be operated by solar PV systems and
AC electric supply mechanisms. The proposed model includes solar panel, DC motor, hand
operated driver & pedal operated driver and a novel pump mechanism. The proposed water
lifting pump is coupled with 350 W, 36 V DC motor. This motor rating is selected for
optimum water output for lower heads of 5-10 meters. Hand driven and pedalling activities
by individuals can results in 20 litres per minute of water. Further, DC motor can be
operated directly from solar panel is hot sunny days during bright sunshine hours. During
lesser radiation conditions, pump can be operated via hand or pedal driven and by grid
connected supply through rectifier circuits. The pumping mechanism is constructed using
PVC pipe structure with water lifting structure along the pipe. The guy rope is located intact
with lifting vanes and this is coupled to the shaft of the wheel connected to prime movers.
The proposed model is tested with all the provision made for operation and the results are
tabulated. The results revealed that the novel pump can be effectively employed for small
scale agriculture. This will significantly save the electricity & diesel dependency and also
provides the sustainable solution for water pumping for small scale agriculture.

Keywords:

Water pump, Centrifugal pump, Pipe Networks, Solar Panels, DC Motor

531
International Journal of Research and Analysis in Science and Engineering

1. Introduction:

Small-scale farming accounts for a large percentage of agriculture in India. A sizable section
of the population in India depends on small-scale agriculture for their livelihood, which is
important to the country's economy and society. Approximately 86% of all operating
holdings are categorized as small and marginal, or smaller than 2 hectares, according to the
Agricultural Census. Fragmented Landholdings and Small & Marginal Farmers are
indicators of Landholding Patterns in India. Due to population growth and inheritance
restrictions, the average landholding size has been declining, resulting in farms that are less
productive and more dispersed. In India, agriculture makes up about 17–18% of the
country's GDP [1-3]. Even though small-scale farming frequently faces low productivity
and income, it remains an essential component. About half of all Indian workers are
employed in agriculture, with small-scale farmers accounting for a sizable share of this
sector. Additionally, they deal with serious problems in the technique of irrigating their little
area covered agriculture fields by using expensive irrigation systems. However, in order to
assist small-scale farmers, the government offers incentives for irrigation equipment. To
boost agricultural activity, efforts are being made to upgrade irrigation systems in rural
areas. Small-scale farmers in India have a practical and sustainable option in hand-held
water pumping systems, especially in areas with shallow water tables [4-8]. They support
small farmers' general way of life by enhancing agricultural yields and streamlining
irrigation techniques. In practice Monoblock and end suction centrifugal pumps pumps are
employed in irrigation systems. The comparative analysis of monoblock and end suction
centrifugal pumps pumps is presented in Table.1. Further, Tabe.2 presents the specifications
and pumping parameters of a submersible pump.

Table 1: Comparison of Monoblock and End Suction Centrifugal Pumps

Parameters Mono block pump End suction centrifugal pumps

Suction size (mm) 50-125 65-100
Delivery size (mm) 40-125 50-100
Impeller diameter (mm) 185-300 109-300
Total head (m) 4-45 6-30
Discharge (l/sec) 1.4-73 4.2-38.0
Revolution per minute 1380-1430 1400-2870
Motor rating (kw) 1.5-7.5 0.75-6.5

Table 2: Specifications of A Submersible Pump

Parameters Specifications
Pump size (mm) 100 150 200 & above
Number of stages 20-30 7-30 3-12
Stage-hp 0.08-0.25 0.25-3.0 3.30-40.0
Total head per stage (m) 1.6-4.6 5.0-10.5 12.0-31.0
Discharge (l/min) 20-350 30-1200 300-650

532
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

In India's agricultural industry, centrifugal pumps are essential since they raise farming
methods' sustainability and productivity. They do, however, have problems with
maintenance and power supplies. Reliance on energy might provide difficulties in places
with erratic power supplies. As a remedy, solar-powered centrifugal pumps are being
investigated. Moreover, durability and effectiveness depend on routine maintenance, which
can be difficult in isolated places [9-14]. In this connection many literatures presented the
novel pumping mechanisms where water pumping is done independent of electricity.

II Background of Water Lifting Mechanisms:

Man's physical power production is limited and may be between 0.08 and 0.1 horsepower.
For irrigation purposes, this power can be used to raise water from shallow depths. The
gadgets that the average person powers are the paddle wheel and swing basket. The swing
basket is a device consisting of a basket with four ropes attached that is built from
inexpensive materials such as iron sheet, leather, or woven bamboo strips. When two people
hold a basket with their backs to each other and dip it into a water source, the basket is lifted
and filled with water from the watercourse where it flows into the fields by swinging. The
gadget is functional down to 15 meters, and its discharge capacity ranges from 3500 to 5000
liters per hour. A paddle wheel is made up of tiny paddles that are radially mounted on a
horizontal shaft. The shaft moves in a tightly fitted concave trough, forcing water in front
of the paddles. The quantity of blades varies based on wheel size; 8 blades for 1.2 meters in
diameter and up to 24 blades for 3 to 3.6 meters. The 12-bladed wheel can raise roughly
18,000 liters per hour from 0.45 to 0.6 meters of water. India had an abundance of animal
power. Along with other field activities and processing jobs, they are utilized for raising
water. Two bullocks could produce about 0.80 horsepower. They are capable of lifting water
up to 30 meters deep. Naturally, as lift increases, the rate of discharge will decrease.

Around 3000 BC, water raising devices have been present in many places of the world. The
energy needed to turn the wheels was provided by animals (muscle energy) in the
construction of early devices like water wheels and chutes. Later on, pumps like the
"Archimedean" helicoid pumps were created and are still in use today. Until the previous
century, a variety of water-lifting machines known as "tympana" (drums) were extensively
employed for mining and irrigation. Shaduf is recognized as the original water-lifting device
in many ancient civilizations. It has been called shaduf (shadoof) in Egypt, zirigum in
Sumer, kilonion or kelonion in Hellas, daliya in Iraq, picottah in Malabar, lat in India, gerani
or geranos in Hellenistic Egypt, kilan (derived from the Greek word kilonion) in Israel, and
tolleno in Latin countries, among other names. It is a manually powered wooden tool for
raising water out of a canal, cistern, river, or well. Its most popular configuration consists
of a long, nearly horizontal wooden pole that taper down to resemble a seesaw. On one end
of the pole, a bag and rope are fastened, and on the other, a counterbalance.

Filling the container, the operator pulls down a rope that is tied to the long end, allowing
the counterweight to hoist the filled container. Sometimes a sequence of shadufs was
arranged one atop the other. 2.5 m3/d was a normal water lifting rate. Thus, 0.1 hectares of
land might be irrigated in 12 hours by a single shaduf. Around 3000 BC, the Mesopotamians
were known to use the shaduf to hoist water. The 18th Dynasty (c. 1570 BC) in Upper Egypt
is when the shaduf, which was already in use in Mesopotamia, first debuted. This was
sometime after 2000 BC.
533
International Journal of Research and Analysis in Science and Engineering

This apparatus made it possible to irrigate farms close to canals and riverbanks during the
year's dry spells. Later on, the system was improved by adding a pulley and animal traction
to raise water levels from deep wells. It is still commonly used today to irrigate small land
plots near wells and to provide drinking water. The Arabian Peninsula also saw adaptations
of the gadget.

III Proposed Mechanism of Novel Pump:

The conceptual block diagram of proposed multi prime mover coupled novel pump is
presented in Figure 1. The system consists of a solar panel, DC motor, water pump, hand
driven and pedal operated driver. Between the solar panel, battery, and DC motor is a charge
controller. Included is a set of rectifiers for feeding DC power from an AC source. The
motor provides mechanical energy to the water pump's rising pulley. With the aid of an
appropriate guide, the rope is pushed by a rising pulley at ground level, which draws it down
into the water tank and up through rising pipe. To prevent piston slippage, pistons are kept
at the ideal distance from one another and equally spaced. This allows water to be raised
and collected in a tank or allowed to flow further as needed. In lower radiation
environments, the water pump is manually operated or pedaled like a stationary bike.

Figure 1: Block diagram of the proposed methodology

The suggested pump is made up of a closed looped rope pushed via a water-immersed
conduit that has pistons attached at equal intervals. A guidance box is installed at the bottom
of it. The main function of the guide box is to raise the rope that descends into the tank and
steer it upward into the riser pipe. The water is raised by the pistons that enter the rising
main pipe until it reaches the top, where there is a spout that allows it to exit. Rotation of
the pulley wheel located at the top will pull the rope down. The water spout is raised by the
piston being pulled through the rising main pipe by the friction between the pulley wheel
and rope. The handle, which is housed in a pump structure atop the tank, doubles as the
pulley wheel axle. A guide located close to the tank's bottom ensures that the rope with the
piston enters the rising main smoothly. The bottom end of the riser pipe's edge prevents
pistons from being hooked. Water is raised and held in the collecting tank. The proposed
water lifting system consists of solar panels, DC motor, Frame, Pulley, Rope, Piston, Rising
pipe, Discharge pipe and a Guide box. Figure 2 presents the conceptual diagram of the
proposed pump. Figure 3 presents the constructed pipe structure to lift the water from the
storage tank.

534
Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

Figure 2: Block Diagram of the Proposed Methodology

Figure 3: Developed Model for Lifting the Water from The Source

535
International Journal of Research and Analysis in Science and Engineering

Figure 4: Solar PV Panels Used for The Pumping

Table 3: Specifications of SPV panels

Parameters Specifications
Model Number HST72F330P
Pmax 330 W
Voc 46.12 V
Isc 9.32 A
Vmp 37.40 V
Imp 8.83 A

Figure 5: DC Motor Coupled to Pump Through Pulley

The torque in a DC motor can be calculated using the motor's electrical parameters. The
equation (1) presents the torque developed by motor.

𝑃𝑒
𝑇= 𝜔
………(1)

536
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

Where,

T is the torque developed by the motor in N-m

Pe is the electrical power in watts

ω is the angular velocity in radians per sec

Figure 6: Connection Diagram with Charge Controller

A pulley, also known as a supporting shell or axle, is a wheel on an axle or shaft used to
facilitate the movement and direction changes of a taut cable. A rope, cable, belt, or chain
that passes over the pulley inside the groove or grooves can serve as the drive element of a
pulley system. To apply significant forces, a block and tackle is constructed from pulleys to
create a mechanical advantage. In belt and chain drives, pulleys are also built to transfer
power from one rotating shaft to another.

Figure 7: Guided Pulley and Wheel Employed in The Model

While they work well for pulling and hoisting, ropes lack the compressive strength. They
are therefore unsuitable for compressive applications such as pushing. Compared to cable,
line, string, and twine that are formed similarly, rope is stronger and thicker. While wire
rope is composed of wire, fiber rope is made of fiber. A rope's qualities include strength,
durability, resistance to water, and non-stretching while use.
537
International Journal of Research and Analysis in Science and Engineering

Additionally, it is not overly smooth to prevent wheel slipping. Rubber can be used to make
pistons; for example, the side section of an old vehicle tire can be used. Wood and leather
have also been used, though less successfully. High-density polyethylene pistons are
effective and simple to manufacture in standard sizes. These ropes are not too long or too
slack. Nylon ropes are an option, but they have a propensity to sag and slip. The ideal rope
diameter is 6 mm. Knots or melting a piece of rope on both sides of the piston might be used
to secure it to the rope. There is one meter separating the two pistons. There will be sliding
between the pistons and the wheel if there are more pistons on the rope. Piston diameter
varies with the size of the rising main. To secure the piston a knot is placed before and after
each piston.

Figure 8: Piston with Rope

A vertical pipe that rises from the earth to feed water to the discharge pipe is called a rising
pipe in this context. The tubes utilized here are low pressure types since the pressure in the
rising main is low. The tube diameter at the top (the discharge and outlet) should be greater
than the diameter of the ascending main tube. The pump structure or another fixed place
should be where the discharge tube is fastened. The volume of water that may be raised is
determined by the rising pipe's diameter; greater diameters allow for the lifting of heavier
water. Thus, in order to avoid making the lifting excessively taxing, smaller diameter pipes
must be used for very small wells. The pipe used to transfer fluid to the exit tank is referred
as the discharge pipe. The discharge pipe is slightly angled in relation to the rising pipe to
enhance fluid flow. The rope and pistons, which fit into the rising main with a 1 mm
clearance, raise the water. The ascent major culminates with a to guarantee a maximum rate
of discharge, the inclination is set at an angle of less than 90°. Another essential component
of the pump is the guide. Its purpose is to prevent the pistons and rope from rubbing against
the raising main's entry by guiding them into it. Using the guidance to determine the ideal
blend of durable materials beneath the water rope—whether it be glass or glaze—is crucial.

Table 4: Proposed Systems Components and Specifications

System
Sl. Specifications
Components
1. Piston Diameter – 35 mm, Thickness - 4 to 5 mm, Material - HDPE
(High density poly ethylene)
2. Rope Diameter – 6 mm, Material- Fiber

538
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

System
Sl. Specifications
Components
3. Guide Pulley Outer & Inner diameter – 65 mm & 22 mm, Width – 70 mm,
Material: Nylon
4. Rising pulley Diameter – 66 cm, Material – Steel
5. Rising and Discharge Diameter – 3.8 cm, Material – PVC
Pipe
6. Guide box Coupler - 4 inch, Eccentric Reducer – 110 X 75 mm, Material
- PVC
7. DC Motor 24 V, 350W
8. Battery 26 Ah, 12 V Lead Acid Exide
9. Solar Panels 250 W, 12 V polycrystalline

Figure 9: Guide Box Employed with Rope and Pulley

Figure 10: Assembled Model of the Water Lifting Pump

539
International Journal of Research and Analysis in Science and Engineering

Figure 11: Hand Held Operation and Water Output of Pump

Figure 12: Side View of the Assembled Model with Peddling Mechanism

Figure 13: Peddling Operation and Water Output of Pump

540
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

IV. Results and Discussions:

The proposed pump is tested for its performance under different conditions. DC motor
coupled to the pump, is tested under different solar radiations. It is observed that with 1000
W/m2 radiation water output was around 50 liters per minute and with decrease in solar
radiation the performance was drastically reduced to 35 liters per minute with 700 W/m2
radiation. Further performance testing with hand driven and pedal driven operations were
conducted. Four individual persons of different capabilities were made to pedal the water
pump and it was observed that the water output per minute was in the range of 20 liters to
14 liters per minute. The performance details with pedaling operation is listed in Table.5.

Table 5: Results of Performance Testing by Hand and Pedal Driven Operations

Sl. Weight (Kg) Avg. Pedaling Speed (RPM) Discharge Lit./min.

Person 1 68 83 20
Person 2 60 70 18
Person 3 55 72 18
Person 4 48 60 14

The suggested pump's working concept makes it possible to create an energy-efficient,

economical, and low-energy water pumping system that meets irrigation needs with both
renewable and human energy sources. There are relatively little radial stresses and static
pressures in the pump pipe because the weight of the water column is evenly distributed
over the pistons. The primary features of this pump are that its output is based on three
factors: the crucial pump speed at which it begins to pump, the cross-sectional area of the
pump pipe, and the rope speed. This pump is somewhat susceptible to rust and silt. It creates
a constant, smooth flow without subjecting the pump pipe or rope to dynamic loading. These
features lead to the conclusion that, when compared to centrifugal and piston pumps, the
novel pump is a more efficient pumping mechanism.

V. Conclusions:

A new concept of water pumping for irrigation systems is presented in the paper. It is
revealed from the experiments that the proposed model can be operated by solar PV, electric
supply and by hand-pedal driven operations. Hand driven and pedaling activities by
individuals resulted in 20 liters per minute of water. The proposed model is tested with all
the provision made for operation and the results are tabulated. The results revealed that the
novel pump can be effectively employed for small scale agriculture. This will significantly
saves the electricity & diesel dependency and also provides the sustainable solution for
water pumping for small scale agriculture. The operating principle of the novel pump
enables a light-weight, cost-effective device for water supply and irrigation that can be used
by families and small communities. It can be produced with locally available materials or
recovered materials. Compared to other hand pumps, the novel pump has a high pumping
capacity and can pump from wells of 1 to 10m deep. If properly produced, installed and
maintained the pump can work efficiently upto 90%. Another great aspect of this model is
that it is ecofriendly.

541
International Journal of Research and Analysis in Science and Engineering

References:

1. Rawat R, Kaushik S, Lamba R, “A review on modeling, design methodology and size

optimization of photovoltaic based water pumping, standalone and grid connected
system” Renewable and Sustainable Energy Reviews, 57:1506–1519, 2016.
2. Islam, S.M., Hossain, M. Z., and Khair, M. A. “Design and Development of Pedal Pump
for Low- Lift Irrigation, Journal of Agriculture and Rural Development (JARD)” vol. 5
(1&2) Pp 116-126, 2007.
3. Waghmare S. N, Mestri M. M, Lavekar P. V, Misal, T. S, and Nalawade P. C.
“Manually Operated Spiral Tube Water Pump” International Research Journal of
Engineering and Technology (IRJET), vol. 2 (1) Pp 167-169, 2015.
4. M. N. Usman, C. C. Mbajiorgu, J. M. VinkingDesign, “Construction and Testing of a
Hand Operated Water Pump for Small Scale Farmers in Nigeria.” International Journal
of Advanced Research in Science, Engineering and Technology Vol. 6, Issue 10,
October 2019.
5. Baumann, E. “Water lifting (1st ed.). St. Gallen, Switzerland: SKAT” Swiss Centre for
Development Cooperation in Technology and Management, 2000.
6. Basanagouda F Ronad, “Design, Simulation and Development of DC-DC Converter for
SPV Powered DC Distribution System in Domestic Loads”, Lecture Notes in Electrical
Engineering, Springer Verlog, ISSN: 1876-1100, Journal No. 4524, March 2022,
pp.777-791.
7. Bartwa A. “Water security through solar PV water pumping. International Journal of
Engineering Science Invention, 7:17–23, 2018.
8. G Pooja, Manjula Gaddi, Rakesh, Sachin Karant, Basanagouda F. Ronad,
“Development of Android Application for Optimization of Irrigation System Capacity”,
Turkish Online Journal of Qualitative Inquiry (TOJQI), ISSN: 1309-6591, Volume 12,
Issue 7, July 2021, pp.- 1034 – 1039.
9. Joshi Sourabha, Basanagouda F. Ronad, Suresh H. Jangamshetti, “Mathematical Model
for Performance Prediction of SPV Powered DC Irrigation System”, TEST Engineering
and Management, ISSN: 0193 – 4120, January - February 2020, pp.- 9000 – 9006.
10. Basanagouda F Ronad, Suresh H Jangamshetti, “A Critical Approach for Energy
Conservation in Irrigation Systems by Optimal Sizing of Pumps”, International Journal
of Innovative Technology and Exploring Engineering, ISSN: 2278-3075, Volume-8,
Issue-6S3, April 2019, pp.-155-159.
11. Raichal Rani, Basanagouda F Ronad, Suresh H Jangamshetti, “Development and
Testing of SPV Powered 1HP AC Water Pumping System”, International Journal of
Scientific & Engineering Research, ISSN 2229-5518, Vol 9, Issue 7, July 2018, pp.-
2045-2049.
12. Basanagouda F Ronad, Suresh H Jangamshetti, “Optimum Sizing of SPV Powered DC
Irrigation Pumps based on Field Conditions”, International Journal of Scientific and
Engineering Research, ISSN- 2229-5518, Vol 7, Issue 10, October 2016, pp.-1738-
1742.
13. V. Hulyal, B. F. Ronad and S. H. Jangamshetti, "An Effective Simulation Model for
Performance Prediction of SPV Powered DC Irrigation Systems-A Case Study for
Validation," 2017 International Conference on Current Trends in Computer, Electrical,
Electronics and Communication (CTCEEC), Mysore, India, 2017, pp. 378-382, doi:
10.1109/CTCEEC.2017.8455171.

542
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

14. B. F. Ronad and S. B. Kumbalavati, "Performance Assessment of SPV Powered DC

Irrigation Pumps with Solar Tracking Mechanism," 2020 3rd International Conference
on Energy, Power and Environment: Towards Clean Energy Technologies, Shillong,
Meghalaya, India, 2021, pp. 1-5, doi: 10.1109/ICEPE50861.2021.9404406.
15. M. Bannur, U. Patil, G. Patil and B. Ronad, "Experimental Investigation of Technical
Feasibility of Standalone Pico-hydro Solar PV Hybrid Power System," 2022 IEEE
North Karnataka Subsection Flagship International Conference (NKCon), Vijaypur,
India, 2022, pp. 1-5, doi: 10.1109/NKCon56289.2022.10126879.
16. B. F. Ronad and S. H. Jangamshetti, "Optimal cost analysis of wind-solar hybrid system
powered AC and DC irrigation pumps using HOMER," 2015 International Conference
on Renewable Energy Research and Applications (ICRERA), Palermo, Italy, 2015, pp.
1038-1042, doi: 10.1109/ICRERA.2015.7418568.
17. Uche, R. Oguoma, O., and Godfrey, E. “Design and Fabrication of a Dual Power Water
Pump, Innovative Systems Design and Engineering” vol. 5, no 6 Pp7-19, 2014.
18. Bhuti, J. Kadakolmath and B. F. Ronad, "Development of Smart Solar Tracker-An
Approach towards Self Governed Tracking Mechanism," 2021 5th International
Conference on Electronics, Communication and Aerospace Technology (ICECA),
Coimbatore, India, 2021, pp. 88-93, doi: 10.1109/ICECA52323.2021.9676000.
19. J. Kumbar, N. Goudar, S. Bhavi, S. Walikar and B. F. Ronad, "Optimization of Hybrid
Renewable Electric Energy System Components for Covid Hospitals," 2021 IEEE
Mysore Sub Section International Conference (MysuruCon), Hassan, India, 2021, pp.
799-804, doi: 10.1109/MysuruCon52639.2021.9641633.
20. V. Biradar, B. F. Ronad and S. H. Jangamshetti, "Development, Testing and
Performance Analysis of SPV Powered 0.5HP AC Water Pumping System," 2017
International Conference on Current Trends in Computer, Electrical, Electronics and
Communication (CTCEEC), Mysore, India, 2017, pp. 353-357, doi:
10.1109/CTCEEC.2017.8454991.
21. M. Naik, S. Kori, B. F. Ronad and S. H. Jangamshetti, "Performance Analysis of
Directly Coupled Solar Photovoltaic Powered Irrigation Pumps," 2018 International
Conference on Recent Innovations in Electrical, Electronics & Communication
Engineering (ICRIEECE), Bhubaneswar, India, 2018, pp. 3414-3419, doi:
10.1109/ICRIEECE44171.2018.9009384.
22. Krishnakant, Kulkarni PS. “Design and simulation of solar photovoltaic water pumping
system” 3rd International Conference on Energy, Power and Environment: Towards
Clean Energy Technologies, Shillong, Meghalaya, India, 5–7 March, 2021.

543
ISSN: 2582-8118 Volume 4, Issue 5; Sept 2024

International Journal of Research and

Analysis in Science and Engineering
Web: https://www.iarj.in/index.php/ijrase/index

Sponsors for SSET 2024

➢ Platinum Sponsor (Rs. 50,000=00)
Mr. Vijay S. Alur
Partner
nd
M/S S. S Alur Constructions, 2 Floor, Vaishanvi Heights, Opposite to KSFC,
Vivek Nagar East Vijayapur 586 109, Karnataka.

544
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

➢ Gold Sponsor (Rs. 25,000=00)

Universal Print Systems
Partner
Plot No. 113, ALEAP Industrial Estate, Surampalli(V),
Gannavaram 521212, Krishna District, Andhra Pradesh.

545
International Journal of Research and Analysis in Science and Engineering

➢ Silver Sponsor (Rs. 10,000=00)

Vasa Scienctic Co
95/1, 11th Cross, 4th Main, Malleshwaram
Bangaluru 560 003.

546
 Design and Implementation of Multi Prime Mover Coupled Novel Water Pump for Small....

➢ Silver Sponsor (Rs. 10,000=00)

Kutub Jalihal Skillzam
1st Floor, Pearl Plaza, Plot #271, Shivabasava Nagar, Nehru Nagar,
Belagavi, Karnataka.

547
International Journal of Research and Analysis in Science and Engineering

➢ Silver Sponsor (Rs. 10,000=00)

Eyesec Cyber Security

➢ Silver Sponsor (Rs. 10,000=00)

Angel Constructions
Bagalkote.

548