Review

A Literature Review on Security in the Internet of Things: Identifying

and Analysing Critical Categories
Hannelore Sebestyen 1, *, Daniela Elena Popescu 2, * and Rodica Doina Zmaranda 2

1 Faculty of Automation and Computing, Politehnica University Timis, oara, 300223 Timişoara, Romania
2 Computers and Information Technology Department, University of Oradea, 410087 Oradea, Romania
* Correspondence: verkman.hanne@gmail.com (H.S.); depopescu@uoradea.ro (D.E.P.)

Abstract: With the proliferation of IoT-based applications, security requirements are be-
coming increasingly stringent. Given the diversity of such systems, selecting the most
appropriate solutions and technologies to address the challenges is a complex activity. This
paper provides an exhaustive evaluation of existing security challenges related to the IoT
domain, analysing studies published between 2021 and 2025. This review explores the
evolving landscape of IoT security, identifying key focus areas, challenges, and proposed
solutions as presented in recent research. Through this analysis, the review categorizes
IoT security efforts into six main areas: emerging technologies (35.2% of studies), securing
identity management (19.3%), attack detection (17.9%), data management and protection
(8.3%), communication and networking (13.8%), and risk management (5.5%). These per-
centages highlight the research community’s focus and indicate areas requiring further
investigation. From leveraging machine learning and blockchain for anomaly detection
and real-time threat response to optimising lightweight algorithms for resource-limited
devices, researchers propose innovative and adaptive solutions to address emerging threats.
The review underscores the integration of advanced technologies to enhance IoT system
security, while also highlighting ongoing challenges. The paper concludes with a synthesis
of security challenges and threats of each identified category, along with their solutions,
aiming to support decision-making during the design approach of IoT-based applications
and to guide future research toward comprehensive and efficient IoT frameworks.

Academic Editor: Paolo Bellavista Keywords: IoT security; attack detection; emergent technologies in IoT; IoT vulnerabilities;
Received: 28 December 2024
adaptive security solutions
Revised: 1 February 2025
Accepted: 8 February 2025
Published: 11 February 2025

Citation: Sebestyen, H.; Popescu,

1. Introduction
D.E.; Zmaranda, R.D. A Literature 1.1. IoT Evolution Overview
Review on Security in the Internet of
The Internet of Things represents one of the most transformative technological ad-
Things: Identifying and Analysing
Critical Categories. Computers 2025, 14,
vancements of the contemporary era. By enabling physical objects to connect to the internet,
61. https://doi.org/10.3390/ exchange data, and interact autonomously, IoT has significantly reshaped various aspects
computers14020061 of our lives, including how we live, work, and communicate. The exponential growth of
Copyright: © 2025 by the authors.
interconnected devices, spanning from everyday household appliances to sophisticated
Licensee MDPI, Basel, Switzerland. industrial machinery, has fostered a highly integrated ecosystem that offers unparalleled
This article is an open access article convenience, efficiency, and potential for innovation [1]. However, this interconnected-
distributed under the terms and ness, while presenting new opportunities for progress, also introduces a series of complex
conditions of the Creative Commons
security challenges that cannot be disregarded.
Attribution (CC BY) license
According to the IoT Analytics platform, it is projected that the number of IoT-
(https://creativecommons.org/
licenses/by/4.0/).
connected devices will surge by 13% annually in 2024, reaching 18.8 billion devices. Figure 1

Computers 2025, 14, 61 https://doi.org/10.3390/computers14020061

 interconnectedness, while presenting new opportunities for progress, also introduces a
series of complex security challenges that cannot be disregarded.
Computers 2025, 14, 61 2 of 45
According to the IoT Analytics platform, it is projected that the number of IoT-con-
nected devices will surge by 13% annually in 2024, reaching 18.8 billion devices. Figure 1
represents a substantial increase of 8.5 billion compared to 2019 [2]. Furthermore, it is an-
represents a substantial increase of 8.5 billion compared to 2019 [2]. Furthermore, it is
ticipated that this number will double by 2030, potentially surpassing 40 billion devices.
anticipated that this number will double by 2030, potentially surpassing 40 billion devices.

Figure 1. Estimated IoT-connected devices (in billions) in the past 10 years [2].
Figure 1. Estimated IoT-connected devices (in billions) in the past 10 years [2].
At its core, the Internet of Things comprises a vast network of physical objects
At
equippedits core,
withthe Internet
sensors, of Things
software, and comprises a vast network
various technologies, enablingof physical
them toobjects
commu-
equipped
nicate withwithone sensors, software,
another and various
and centralised technologies,
systems enabling them
via the internet. to communi-
This communication
cate with onereal-time
facilitates another and datacentralised
collection systems via thedriving
and analysis, internet.intelligent
This communication
automationfacil- across
itates real-time
diverse sectors data collection
including and analysis,
healthcare, driving intelligent
transportation, automation
agriculture, across diverse
energy management, and
sectors including healthcare, transportation, agriculture, energy management,
urban planning. In smart cities, for instance, IoT technologies are used to optimise traffic and urban
planning. In smart
flow, reduce energycities, for instance,
consumption, IoT technologies
enhance public safety, are used
and evento manage
optimisewastetraffic flow,
efficiently.
reduce While
energythese consumption, enhance public safety, and even manage waste
capabilities offer substantial opportunities for enhancing efficiency and efficiently.
While thesethey
convenience, capabilities
also entailoffer a substantial opportunities
critical vulnerability: the for enhancing
security efficiency
of these devices.andThe
convenience,
very attributesthey that
also make
entail IoT
a critical
devicesvulnerability:
appealing—suchthe security
as theofability
these devices.
to collectTheandvery
trans-
attributes that make
mit sensitive IoT devices
data—also render appealing—such
them susceptible as the ability
to cyber to Each
threats. collectadditional
and transmit device
sensitive
connecteddata—also render them
to the internet expands susceptible to cyber
the digital attackthreats.
surface,Each additional
creating moredevice con-en-
potential
nected to thefor
try points internet expands
malicious actors.the A digital attack surface,
compromised devicecreating
can serve more potential
as an entry
entry gateway
points for malicious
for attackers, actors.
enabling A compromised
them device
to infiltrate entire can servesteal
networks, as ansensitive
entry gateway
data, orfor at-
disrupt
tackers,
criticalenabling them to infiltrate entire networks, steal sensitive data, or disrupt critical
infrastructure.
infrastructure.
The urgency of securing the Internet of Things has never been more pressing. As the
The urgency
number of securing
of connected devices the continues
Internet oftoThings
rise, sohas
donever been more
the threats pressing. As the
and vulnerabilities that
number of connected
they introduce. In thedevices
context continues to rise, so doisthe
of IoT, cybersecurity notthreats
merelyand vulnerabilities
about that
protecting devices
they introduce.
from unauthorisedIn theaccess;
contextitof IoT, cybersecurity
encompasses is not merely
safeguarding about protecting
entire ecosystems devices
of interconnected
from unauthorised
systems access;
from a diverse it encompasses
range of cyber threats.safeguarding entire ecosystems
The risks associated of intercon-
with inadequate security
are far-reaching—personal
nected systems from a diversedata range mayof be exposed,
cyber critical
threats. infrastructure
The risks associatedcan be compromised,
with inadequate
and public
security trust in these technologies
are far-reaching—personal datamaymaydiminish.
be exposed,Several high-profile
critical incidents,
infrastructure can besuch
as attacks onand
compromised, unsecured
public trust smart hometechnologies
in these devices, have mayalready underscored
diminish. the potential
Several high-profile
consequences
incidents, such as ofattacks
IoT vulnerabilities,
on unsecured raising
smartalarms
home in both the
devices, public
have and private
already sectors.
underscored
Emerging
the potential technologies
consequences of play a pivotal role inraising
IoT vulnerabilities, enhancing security
alarms in bothmeasures
the publicagainst
andthe
evolving
private landscape of cyber threats within IoT environments. By integrating solutions such
sectors.
as Artificial Intelligence, Blockchain, Machine Learning and other innovative technologies,
organisations can construct more robust defences against sophisticated attacks. For in-
stance, AI-powered anomaly detection systems can assist in identifying unusual patterns
of behaviour within IoT networks, facilitating expedited detection of potential breaches.
Computers 2025, 14, 61 3 of 45

Blockchain, with its decentralised and immutable ledger, provides a means of securing
data exchanges between devices and ensuring the integrity of communications. Public Key
Infrastructure systems can provide enhanced authentication mechanisms for IoT devices,
thereby reducing the likelihood of unauthorised access. As these technologies continue to
evolve, they will play an indispensable role in addressing the unique security challenges
posed by the interconnected nature of IoT.
In the realm of Internet of Things technology, cyber threats are exhibiting a remarkable
level of sophistication. Hackers are increasingly targeting devices equipped with inad-
equate or insufficient security measures, thereby gaining unauthorized access to larger
networks. Factors such as inadequate encryption, the absence of robust authentication
protocols, and outdated software contribute to the surge in cyber incidents within the IoT
ecosystem. Consequently, addressing these vulnerabilities necessitates a comprehensive
and proactive security strategy that transcends mere technical solutions. It encompasses not
only technical measures but also well-defined policy frameworks and industry standards.
To underscore the paramount importance of addressing IoT security vulnerabilities,
Table 1 presents a comprehensive overview of the most significant IoT-related attacks that
occurred between 2015 and 2024.

Table 1. Significant IoT-related attacks 2015–2024.

Year Attack Targeted IoT Domain Process Description Impact

Security researchers remotely controlled Chrysler recalled 1.4M vehicles for
2015 Jeep Cherokee Hack [3,4] Automotive IoT
a Jeep via its IoT-connected systems. security upgrades.
Malware infected IoT devices like
Major websites disrupted; large-scale
2016 Mirai Botnet Attack [4] IoT Consumer Devices routers and cameras, creating a
DDoS attacks.
massive botnet.
Exploited unpatched systems in
$4 billion in damages globally; disrupted
2017 WannaCry Ransomware [5] Industrial IoT IoT-connected healthcare devices
hospitals and critical infrastructure.
and networks.
Attackers used an IoT-connected
Casino IoT Thermometer Sensitive customer data stolen;
2018 Smart Aquarium thermometer to access a casino’s
Hack [6] significant reputational damage.
high-roller database.
Hackers accessed poorly secured Ring
Privacy violations; public outcry over
2019 Ring Doorbell Hacks [7] Consumer IoT Devices IoT cameras, spying on and
security flaws.
harassing users.
Ransomware disabled Garmin’s
Garmin Ransomware Multi-day outage; $10M ransom
2020 IoT Fitness Devices IoT-connected services, including
Attack [8] reportedly paid.
aviation and fitness.
Hackers exploited compromised
Colonial Pipeline Shutdown of pipeline; $4.4M ransom
2021 Energy Infrastructure credentials to access pipeline’s
Ransomware [9] paid; fuel shortages.
IoT-linked systems.
Attackers accessed 150,000 IoT cameras Exposure of videos from Tesla, hospitals,
2021 Verkada Camera Hack [10,11] IoT Surveillance Cameras
due to exposed admin credentials. and jails.
Hackers attempted to change chemical
Oldsmar Water Treatment Potential public health threat; system
2023 Public Utilities levels in drinking water via IoT
Attack [12] restored quickly.
SCADA systems.
Exploitation of a zero-day vulnerability Data of millions exposed; over $100M in
2023–2024 MOVEit Data Breach [13] Managed File Transfer Tool
in IoT-adjacent systems. regulatory fines/penalties.

The analysis presented in Table 1 underscores several critical observations regarding

the nature and implications of security attacks on IoT systems that have occurred over the
past decade:
1. Diversity of attack domains
IoT systems across a broad range of domains have been targeted, reflecting the exten-
sive integration of IoT technologies in both consumer and industrial sectors. Attacks
on consumer devices, including wearables and smart home systems (e.g., Mirai Botnet,
Ring Doorbell Hacks, Garmin Ransomware), highlight the vulnerabilities inherent in
devices used daily by individuals. Similarly, industrial systems (e.g., the Jeep Chero-
kee Hack) and critical infrastructure (e.g., Colonial Pipeline Ransomware, Oldsmar
 Attacks on consumer devices, including wearables and smart home systems (e.g.,
Mirai Botnet, Ring Doorbell Hacks, Garmin Ransomware), highlight the vulnerabili-
ties inherent in devices used daily by individuals. Similarly, industrial systems (e.g.,
Computers 2025, 14, 61 the Jeep Cherokee Hack) and critical infrastructure (e.g., Colonial Pipeline Ransom- 4 of 45
ware, Oldsmar Water Treatment Attack) have been compromised, emphasizing the
risks to operational continuity, public safety, and essential services.
Water Treatment Attack) have been compromised, emphasizing the risks to operational
2. Economic and social impact
continuity, public safety, and essential services.
The financial and operational consequences of IoT-related attacks have been pro-
2. Economic and social impact
found. High-profile incidents such as the Garmin Ransomware, WannaCry Ransom-
The financial and operational consequences of IoT-related attacks have been profound.
ware, and Colonial Pipeline Ransomware illustrate the significant economic losses
High-profile incidents such as the Garmin Ransomware, WannaCry Ransomware,
incurred through ransom payments, downtime, and operational disruptions. These
and Colonial Pipeline Ransomware illustrate the significant economic losses incurred
attacks also underscore the social ramifications, including the erosion of public trust,
through ransom payments, downtime, and operational disruptions. These attacks also
exposure of sensitive personal and organizational data, and heightened concerns re-
underscore the social ramifications, including the erosion of public trust, exposure of
garding the reliability and security of IoT-enabled systems. For instance, breaches of
sensitive personal and organizational data, and heightened concerns regarding the
consumer devices like Ring cameras not only caused privacy violations but also in-
reliability and security of IoT-enabled systems. For instance, breaches of consumer
stilled a sense of insecurity among users regarding the safety of their connected en-
devices like Ring cameras not only caused privacy violations but also instilled a sense
vironments.
of insecurity among users regarding the safety of their connected environments.
3. Evolving threat landscape
3. Evolving threat landscape
Over the past decade, the sophistication of IoT-related cyberattacks has escalated
Over the past decade, the sophistication of IoT-related cyberattacks has escalated
markedly. Early attacks, such as the Mirai Botnet, exploited relatively simple vulner-
markedly. Early attacks, such as the Mirai Botnet, exploited relatively simple vul-
abilities like default credentials and unsecured interfaces. However, more recent in-
nerabilities like default credentials and unsecured interfaces. However, more recent
cidents, including the MOVEit Data Breach, demonstrate the increasing prevalence
incidents, including the MOVEit Data Breach, demonstrate the increasing prevalence
of zero-day exploits and advanced, targeted attacks. This evolution highlights the
of zero-day exploits and advanced, targeted attacks. This evolution highlights the
growing technical capabilities of attackers and underscores the urgent need for ro-
growing technical capabilities of attackers and underscores the urgent need for robust
bust security measures and proactive defense mechanisms in IoT ecosystems.
security measures and proactive defense mechanisms in IoT ecosystems.
In support of Table 1, a recent statistic published by the Statista website highlights
In support of Table 1, a recent statistic published by the Statista website highlights
the substantial increase in malware attacks targeting IoT systems. As depicted in Figure
the substantial increase in malware attacks targeting IoT systems. As depicted in Figure 2,
2, the number of such attacks has surged nearly fourfold over the past five years, surpas-
the number of such attacks has surged nearly fourfold over the past five years, surpassing
sing 112 million incidents in 2022. This significant growth can be attributed to two primary
112 million incidents in 2022. This significant growth can be attributed to two primary
factors: firstly, the rapid proliferation of IoT devices (as illustrated in Figure 1), and sec-
factors: firstly, the rapid proliferation of IoT devices (as illustrated in Figure 1), and secondly,
ondly, the diverse range of domains in which these devices are deployed, which conse-
the diverse range of domains in which these devices are deployed, which consequently
quently enhances their appeal to malicious actors.
enhances their appeal to malicious actors.

Figure 2. Estimated annual number of IoT malware attacks (in millions) 2018–2022 [14].
Figure 2. Estimated annual number of IoT malware attacks (in millions) 2018–2022 [14].
Given the escalating prevalence of IoT devices and the corresponding surge in cyberat-
tacks, it is imperative that effective solutions be developed to safeguard both the technology
itself and its users.

1.2. Regulatory Overview

Considering the escalating risks associated with IoT devices, regulatory authorities
have implemented stringent security standards. In the European Union, these standards
Computers 2025, 14, 61 5 of 45

are codified in the 2016 General Data Protection Regulation [15] issued by the European
Parliament, which was enacted in response to technological advancements and global inte-
gration. The United States issued the IoT Cybersecurity Improvement Act of 2020, calling
for the National Institute of Standards and Technology and the Office of Management
and Budget to develop standards that establish minimum requirements and guidelines for
the management of IoT devices owned by federal agencies, i.e., the proper management
of information held by them [16]. These standards will necessitate periodic reviews and
updates by NIST every five years.
Over the years, there have been groups and organisations that, noticing the need for
increased cybersecurity, have developed frameworks and standards for different domains.
Thus, in 2017 the IoT Cybersecurity Alliance was formed, consisting of the firms AT&T,
IBM, Nokia, Palo Alto Networks, Symantec, and Trustonic, with the objective of solving
the main cybersecurity challenges in the IoT ecosystem using the expertise of the firms
involved [17]. Another such grouping is the Industry IoT Consortium, which is active in
the industry domain. They developed the first version of the Industry Internet of Things
Security Framework in 2016 with the aim of securing ICS/SCADA systems [18]. It provides
proposals for architectures that can be used and a set of best practices.
The Internet Engineering Task Force is developing standards for providing secure
communication protocols. One such protocol is CoAP—RFC 7252 Constrained Application
Protocol, developed for resource-constrained networks within the IoT ecosystem. It uses
DTLS to secure data exchange [19].
The Organization for Standardization published ISO/IEC 30141, republished in 2024,
which helps in the design of IoT ecosystems by providing best practices for authentication,
data security, and network integrity [20]. Other standards related to cybersecurity in the
IoT ecosystem are those representing the NIST 8259 series, developed by the NIST, and EN
303 645 [21], developed by the European Telecommunications Standards Institute.
Despite the existence of these standards, guidelines, and frameworks, IoT vulnerabili-
ties are continuously present, and the spread of IoT increases the need for solutions.

1.3. Previous Reviews and Our Work

In recent years, organizations have intensified their efforts to regulate the IoT domain,
while researchers have also demonstrated an increasing interest in identifying the most
suitable frameworks to mitigate cyber threats and prevent potential attacks. Despite
significant progress, a definitive solution has yet to be established, prompting ongoing
research in this field.
To contribute to these efforts and provide a comprehensive guide to the theoretical
foundations and existing vulnerabilities, this review systematically analyses recent ad-
vancements in IoT security. While numerous studies have already examined cybersecurity
in IoT systems—some focusing on specific IoT domains, others addressing particular sys-
tem vulnerabilities, and some evaluating the overall evolution of security research—our
objective is to determine the added value that this review can bring to the field.
To achieve this, we conducted an extensive comparative analysis of review papers pub-
lished in the past two years (2023–2024). Table 2 presents a structured comparison of these
studies, utilizing the six key categories identified in our research: attack detection, data
management and protection, securing identity management, communication and network-
ing, emerging technologies, and risk management. The selection criteria for the reviewed
studies prioritized thematic relevance and alignment with the core focus of our paper,
ensuring a meaningful comparison of contributions within the IoT security landscape.
Computers 2025, 14, 61 6 of 45

Table 2. Comparative analysis of recent IoT security review papers, categorized by key security
focus areas.

Attack Data Management Securing Identity Communication Emerging

Review Risk Management Domain
Detection and Protection Management and Networking Technologies
√ √ √ √ √ √ √
Our Work
√ √ √
[22] - partially - General
√ √
[23] - - - - General
√ √ √
[24] - - - General
√
[25] - - - partially - Consumer
√ √
[26] - - - - Smart Homes
√
[27] - - - partially - General
√
[28] - - - partially - General
√ √
[29] - - - - Smart cities
√
[30] - - - partially - General
√ √
[31] - - - - General
√
[32] - - - partially - General
√
[33] - - - partially - General
√
[34] - - - partially - General
√
[35] - - - partially - ICN-IoT
√ √
[36] - - - - General
√ √ √ √
[37] partially - General
√ √
[38] - - partially - General
√ √
[39] - - partially - IIoT
√
[40] - - - partially - General
√
[41] - - - partially - General
√
[42] - - - - - IoMT
√ √ √ √
[43] - - General
√ √ √
[44] - - - General
√ √
[45] - - - - General
√ √ √
[46] - partially - General
√ √ Resource-
[47] - - - -
constrained
√
[48] - - - - - General
√
[49] - - - - - General
√ √ √ √
[50] partially - General
√
[51] - - - partially - General
√ √
[52] - - - - General
√
indicates the presence of a discussion about the category. - denotes the absence of a discussion.

The comparative analysis presented in Table 2 highlights significant trends and gaps
in recent IoT security reviews. A key observation is that while numerous studies have
addressed specific aspects of IoT security, very few provide a comprehensive perspec-
tive encompassing all critical dimensions. In contrast, the present work systematically
examines six fundamental security categories offering a holistic synthesis of challenges,
advancements, and potential solutions.
One of the most striking findings is the lack of emphasis on risk management across
existing reviews. Risk management plays a pivotal role in IoT security, influencing threat
modeling, mitigation strategies, and resilience planning. However, as the table demon-
strates, only one other review [43] explicitly considers this dimension. This underscores a
significant research gap, which our work seeks to bridge by integrating a structured analysis
of risk assessment frameworks and security assurance strategies within IoT ecosystems.
While emerging technologies such as Artificial Intelligence, Blockchain, Machine
Learning, and Edge Computing are frequently referenced in the literature, their practical
integration into IoT security frameworks remains underexplored or only partially addressed
Computers 2025, 14, 61 7 of 45

in most prior reviews. Existing studies often examine these technologies individually, rather
than considering how they could be strategically integrated to enhance IoT security in a
comprehensive manner. While our review does not propose a unified framework combining
these technologies, it provides a systematic analysis of their applications, advantages, and
limitations. By doing so, this work identifies key research gaps and highlights the need
for future studies to explore how these technologies could be effectively combined into
cohesive security architectures that better address IoT-specific challenges.
While some studies concentrate on specific IoT branches or device categories, allowing
for a more in-depth analysis of their central topics, the present study adopts a broader
perspective. It aims to provide a comprehensive overview of IoT security, encompassing its
vulnerabilities, potential solutions, and existing challenges, to serve as both a starting point
and a holistic perspective for researchers and practitioners.
This review seeks to offer a clear synthesis of IoT security challenges and solutions,
serving as a foundational guide for developing resilient and secure IoT systems. Within
this context, the study systematically analyses the primary security concerns in IoT envi-
ronments, with its key contributions being the following:
• Identification of critical security weaknesses frequently addressed in IoT research.
• Examination of the specific difficulties involved in securing IoT devices.
• Review and evaluation of existing solutions designed to mitigate IoT-related secu-
rity risks.
• Analysis of key trends, best practices, and emerging technologies, including Artificial
Intelligence, Blockchain, Machine Learning, and Edge Computing, which are shaping
the future of IoT security.
• Emphasis on the need for robust and comprehensive security strategies to protect
sensitive data and strengthen public trust in IoT technologies.
In contrast to previous reviews that concentrate on specific aspects of IoT security,
this study offers a comprehensive evaluation of IoT security challenges, encompassing six
critical categories. Additionally, our work integrates research conducted between 2021 and
2024, ensuring that the assessment is current and up-to-date with the latest security trends
and advancements. By addressing these gaps, this review serves as a more comprehensive
and actionable resource, supporting researchers and practitioners in designing secure and
resilient IoT systems.

1.4. IoT Architectural Overview

The IoT infrastructure relies on a multitude of interconnected components dis-
tributed across various levels of the system, collectively forming its architecture [53].
While there is no universally standardized architecture for IoT systems, the three-layer
architecture—comprising the application, network, and perception layers—is the most
referenced in the reviewed studies (Figure 3).
Each of these layers performs distinct functions, enabled by specific tools and tech-
nologies. The Perception Layer operates directly with the physical world, gathering data
from its environment and taking necessary actions. Devices at this level are designed
for sensing, data collection, and direct interaction with the external environment when
required [54]. Examples of devices in this category include sensors, actuators, and other
resource-constrained devices [53].
At an architectural level, the performance of these resource-constrained devices could
be optimised using switching controllers. Switching controllers are control systems de-
signed to manage the operation of IoT devices by dynamically enabling or disabling
specific functionalities based on the devices’ operational context. For instance, when a
sensor is not actively collecting data (Deep Sleep Mode) [55], a switching controller can
 Computers 2025, 14, 61 8 of 45

turn off or reduce the power to certain components, effectively conserving energy. This
approach is particularly advantageous in IoT systems where devices are often deployed
in remote or resource-limited environments and rely on batteries or intermittent energy
sources. By using switching controllers, energy consumption can be minimised, device lifes-
pan extended, and overall system efficiency improved [56]. Integrated on/off controllers
Computers 2025, 14, x FOR PEER REVIEW 8 of 47
can disconnect circuits entirely from power sources when not in use, further enhancing
energy conservation.

Figure3.3.Three-layered
Figure Three-layeredIoTIoT
system architecture
system [35]. [35].
architecture

Another strategy or a complementary one is including dynamic clock reconfiguration.

Each of these layers performs distinct functions, enabled by specific tools and tech
Dynamic clock reconfiguration allows devices to adjust their clock frequency based on
nologies. The Perception Layer operates directly with the physical world, gathering data
processing requirements, significantly reducing power usage during idle or low-activity
from its[57].
periods environment and taking necessary actions. Devices at this level are designed for
sensing,
Thesedata collection,
technologies areand direct interaction
particularly beneficial inwith the external
scenarios environment
where large-scale when re
IoT de-
quired [54].
ployments Examples
require of devices
consistent in thiswhile
performance category include
adhering sensors,
to strict energyactuators, and other re
constraints.
source-constrained
The Network Layer devices [53].data transfer between the Perception Layer and the
ensures
Application Layer. This layer encompasses
At an architectural level, the performance gateway devices responsible
of these for aggregating, devices
resource-constrained
storing,
could be and directing data
optimised using to switching
cloud platforms. These devices
controllers. facilitate
Switching communication
controllers with systems
are control
resource-constrained devices using low-power protocols while also interfacing
designed to manage the operation of IoT devices by dynamically enabling or disabling with cloud
servers via robust communication protocols [53]. Depending on the connected devices,
specific functionalities based on the devices’ operational context. For instance, when a
coverage area, and data volume, various types of wireless networks can be employed to
sensor is not actively collecting data (Deep Sleep Mode) [55], a switching controller can
establish these connections [58]:
turn off or reduce the power to certain components, effectively conserving energy. This
• Cellular connections utilizing LPWAN, such as LTE-M and NB-IoT standards, as well
approach is particularly advantageous in IoT systems where devices are often deployed
as unlicensed solutions like LoRa and Sigfox;
in
• remote or resource-limited
Local and environments
personal area networks, and rely
including Wi-Fi and on batteries or intermittent energy
Bluetooth;
sources.
• Mesh protocols, with Zigbee and RFID being the most common. can be minimised, device
By using switching controllers, energy consumption
lifespan extended, and
The Application overall
Layer deliverssystem efficiency
services improved
to end-users [56].
via mobile andIntegrated on/off control
web applications.
lers
A can disconnect
common example iscircuits entirely which
cloud platforms, from power
process sources
collectedwhen notpresent
data and in use,itfurther
to users enhanc
ing energy
through conservation.
dashboards or control functions.
Another strategy or a complementary one is including dynamic clock reconfigura
2.
tion. DynamicMethodology
Research and Paper
clock reconfiguration Structure
allows devices to adjust their clock frequency based
The methodology used in this review was designed
on processing requirements, significantly reducing power to ensure a comprehensive,
usage during idle orsys-low-activ
tematic, and [57].
ity periods reproducible evaluation of the literature on IoT security. It encompasses three
These technologies are particularly beneficial in scenarios where large-scale IoT de
ployments require consistent performance while adhering to strict energy constraints.
The Network Layer ensures data transfer between the Perception Layer and the Ap
plication Layer. This layer encompasses gateway devices responsible for aggregating
Computers 2025, 14, 61 9 of 45

key stages: identification, screening, and eligibility assessment, guided by the PRISMA
(Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework.

2.1. Selection of Article Sources

The articles underlying this review are open access, allowing interested parties to
analyse them and, with their help, to find the best solutions for IoT security. The sources
consulted were chosen based on their reputation, accessibility, and relevance to the field.
The sources of articles were as follows:
• MDPI, A robust platform that encourages scientific exchange and provides a vast
database of articles, offering advanced search capabilities using keywords and topics;
• IEEE Xplore, a comprehensive digital library providing access to a wide range of
technical literature in engineering, computer science, and related fields;
• Cornell University Arxiv, an open-access repository of preprints spanning multiple
disciplines, including computer science and cybersecurity;
• Informatics in Education, which provides access to educational and research-focused
papers in informatics;
• Elsevier, which provides a wide range of services, including access to a vast collection
of academic journals, books, and research databases;
• Springer, a platform that provides access to scholarly articles and books on a variety
of topics, including advanced technologies and IoT security;
• Other sources, including Nature, Informatics in Education, Acadlore, Migration Let-
ters, and Sciendo, each providing valuable contributions to academic research, open-
access publishing, and interdisciplinary studies across diverse fields.
These sources collectively ensure comprehensive coverage of the topic, allowing for a
diverse range of perspectives and insights to be included in the review.

2.2. Search Method

In the MDPI database, the search bar was used to locate articles by title and keywords.
The keywords employed included “IoT security”, “IoT systems”, “IoT communication”,
“IoT vulnerabilities”, “IoT security risk management” and “6G network IoT”. This keyword-
based search returned hundreds of articles that, to varying degrees, address the topic of
security within the IoT ecosystem. To narrow down the results to the most relevant and
up-to-date studies, a publication date filter was applied, restricting the selection to articles
published between 2021 and 2025, mainly 2023–2024. Additionally, a subject-area filter was
used, focusing on engineering, computer science, and mathematics.
This approach ensured the inclusion of recent, high-quality studies that align with the
technical focus of this review while eliminating outdated or less relevant content.
When an article was identified as part of a specific issue, the entire issue was examined
to uncover additional articles connected to the original topic. This approach aimed to
deepen the exploration and identify alternative or complementary solutions. The categories
of the selected articles from these issues are as follows:
• Machine Learning for Cybersecurity: Threat Detection and Mitigation;
• Network Security in Artificial Intelligence Systems;
• Data Security Approaches for Autonomous Systems, IoT, and Smart Sensing Systems;
• Advanced 5G and beyond Networks;
• Key Enabling Technologies for Beyond 5G Network;
• Advances in Internet of Things Technologies and Cybersecurity.
In the other sources, only the search bar was used with the above-mentioned keywords
completed with “Generative AI” and “Digital Identity”.
Computers 2025, 14, 61 10 of 45

2.3. Articles Selection Method

2.3.1. Identification and Screening
In the identification stage of selection, articles were identified through a comprehensive
search across the above-mentioned sources using predefined keywords related to IoT
security. The initial dataset included 971 articles identified across all sources, which were
documented in a Google Sheet for streamlined management. Duplicate records (23) were
removed, resulting in 948 articles for screening. Titles and abstracts were reviewed to
ensure relevance to IoT security, and articles addressing unrelated domains were excluded
(501). The data extracted for each article included the title, keywords, abstract, conclusions,
challenges, and proposed solutions. For reference management, the Mendeley application
was used.

2.3.2. Eligibility
During the eligibility phase, 447 articles underwent a more detailed evaluation. In-
troductory sections, tables, diagrams, and conclusions were reviewed to ensure relevance.
Articles were excluded if any of the following conditions applied:
• The primary focus diverged from IoT security;
• They were editorials, opinion pieces, or predominantly literature reviews without new
solutions or insights;
• They lacked a clearly defined or described solution, framework, or implementation
related to IoT security.
Finally, 95 articles met the inclusion criteria and were included in the review. These
articles presented original solutions to IoT-specific security challenges, with clear method-
ologies and rigorously supported findings.

2.3.3. Evaluation of Methodological Rigor

To enhance the evaluation of articles, a methodological rigor checklist was employed,
assessing the following criteria:
1. Are they explicitly stated, well-defined, and aligned with IoT security challenges?
2. Are the chosen research methods appropriate for addressing the defined objectives?
Do they follow established IoT security research frameworks?
3. Are the techniques sufficiently detailed, transparent, and reproducible? Are statistical
analyses validated?
4. Does the study propose novel insights, frameworks, or technological advancements?
Only articles meeting these criteria were included in the final dataset.
In the PRISMA Flow below (Figure 4), the paper selection procedure can be seen.
The PRISMA flow chart illustrates the detailed selection procedure:
• Identification—Articles were retrieved from MDPI (601 articles), Springer (72 articles),
IEEE Xplore (65 articles), Elsevier (218 articles), Arxiv (5 articles), or Other (10 articles);
• Screening—Articles irrelevant to IoT security were excluded after title, keywords,
abstract, and conclusion reviews;
• Eligibility—Articles lacking methodological rigor or well-defined solutions were ex-
cluded during detailed analysis.
The iterative categorization process ensured that articles addressing multiple IoT
vulnerabilities were allocated to all relevant categories for comprehensive coverage. This
process refined the initial categories into six critical areas of IoT security: attack detection,
data management, securing identity management, communication, emergent technologies,
and risk management.
 Computers
Computers 14, x 14,
2025,2025, FOR61PEER REVIEW 11 of 11
47 of 45

Articles extracted from

databases (n = 971)

Identification
MDPI (n = 601)
Springer (n = 72) Duplicate articles
IEEEXplore (n = 65) (n = 23)
Elsevier (n = 218)
Arxiv (n = 5)
Other (n = 10)

Screened articles Articles excluded

(n = 948) (n = 501)
Screening

Articles assessed eligible Articles excluded

(n = 447) (n = 352)
Eligibility

Final articles selected for

the review
(n = 95)

Figure
Figure 4. PRISMA
4. PRISMA Flow—the
Flow—the selection
selection procedure.
procedure.

The This
PRISMAmethodological approach aimed
flow chart illustrates to provide
the detailed a robust
selection foundation for synthesizing
procedure:
security challenges and solutions, ensuring that the findings of this review reflect the
• Identification—Articles were retrieved from MDPI (601 articles), Springer (72 arti-
Computers 2025, 14, x FOR PEER REVIEW 12 of 47
diversity and complexity of IoT security literature.
cles), IEEE Xplore (65 articles), Elsevier (218 articles), Arxiv (5 articles), or Other (10
In Figure 5, the distribution of articles by their source can be observed.
articles);
• Screening—Articles irrelevant to IoT security were excluded after title, keywords,
abstract, and conclusion reviews;
• Eligibility—Articles lacking methodological rigor or well-defined solutions were ex-
cluded during detailed analysis.
The iterative categorization process ensured that articles addressing multiple IoT
vulnerabilities were allocated to all relevant categories for comprehensive coverage. This
process refined the initial categories into six critical areas of IoT security: attack detection,
data management, securing identity management, communication, emergent technolo-
gies, and risk management.
This methodological approach aimed to provide a robust foundation for synthesizing
security challenges and solutions, ensuring that the findings of this review reflect the di-
versity and complexity of IoT security literature.
In Figure 5, the distribution of articles by their source can be observed.
Numberofofarticles
Figure5.5.Number
Figure articlesby
bysources.
sources.

The literature review in this study was primarily based on publications from MDPI
due to its robust search engine, extensive journal collection, and rapidly updated data-
base. These features facilitated efficient access to high-quality, peer-reviewed articles
across a broad spectrum of topics relevant to IoT systems. While MDPI provided a reliable
 Figure 5. Number of articles by sources.

Computers 2025, 14, 61 12 of 45

The literature review in this study was primarily based on publications from MDPI
due to its robust search engine, extensive journal collection, and rapidly updated data-
base. These features facilitated efficient access to high-quality, peer-reviewed articles
The literature review in this study was primarily based on publications from MDPI
across a broad spectrum of topics relevant to IoT systems. While MDPI provided a reliable
due to its robust search engine, extensive journal collection, and rapidly updated database.
and comprehensive foundation for this review, we acknowledge the importance of diver-
These features facilitated efficient access to high-quality, peer-reviewed articles across a
sifying sources to minimise potential biases and ensure a holistic representation of the
broad spectrum of topics relevant to IoT systems. While MDPI provided a reliable and
field.
comprehensive foundation for this review, we acknowledge the importance of diversifying
While Figure 5 highlights the distribution of IoT security articles based on their
sources to minimise potential biases and ensure a holistic representation of the field.
sources, it is equally important to examine how these publications are distributed across
While Figure 5 highlights the distribution of IoT security articles based on their sources,
academic journals. Considering the substantial number of articles retrieved from the
it is equally important to examine how these publications are distributed across academic
MDPI database, it was found necessary to include a chart illustrating the distribution of
journals. Considering the substantial number of articles retrieved from the MDPI database,
these articles across the various journals in which they were published. This perspective
it was found necessary to include a chart illustrating the distribution of these articles across
provides deeper insight into the scholarly focus and key contributors to the field of IoT
the various journals in which they were published. This perspective provides deeper insight
security research. Figure 6 illustrates the breakdown of articles by journal, shedding light
into the scholarly focus and key contributors to the field of IoT security research. Figure 6
on which publications are at the forefront of disseminating knowledge in this rapidly
illustrates the breakdown of articles by journal, shedding light on which publications are at
evolving domain.
the forefront of disseminating knowledge in this rapidly evolving domain.

Figure 6. Articles from MDPI by journals.

Figure 6. Articles from MDPI by journals.
The following part of this paper is organized into four main sections. The first section
identifies and categorizes the principal types of vulnerabilities discussed in the analysed
literature, offering a comprehensive overview of IoT attack vectors. The second section
explores the challenges associated with these vulnerabilities, examining proposed solutions
such as frameworks, methodologies, and mechanisms for attack detection and preven-
tion. Additionally, this section addresses strategies designed to secure sensitive user data
and protect privacy, reflecting the increasing importance of safeguarding information in
IoT ecosystems.
The discussion section highlights key areas of active research and identifies unresolved
challenges that warrant future exploration. Finally, the conclusion synthesizes the key
insights derived from this review and proposes potential future research directions for each
identified category.

3. Category Identification and Analysis

By analysing the current state of the art from the articles subject to this review, Table 3
realizes a classification of them according to the methodology in the field of IoT systems.
For each category, a subclassification of the targeted issues of related articles was identified.
Computers 2025, 14, 61 13 of 45

Table 3. Category identification and targeted issues.

Categories Related Challenges Targeted Issues References

Intrusion and
anomaly detection;
DDoS attacks;
Increasing number of cyberattacks on IoT devices,
Attack detection Eavesdropping attacks; [35,59–83]
difficulty in detecting attacks in real time.
Concept drift detection and adaptation;
Botnet detection;
Cyberattacks
Data security;
Data privacy;
Vulnerabilities in the storage and transfer of
Data management and protection Digital Identity and Identity-based [84–95]
sensitive data, privacy risks.
encryption;
Generative AI
Authentication of users and devices, Device identification;
Securing identity management [73,85,87,88,91,96–118]
management of unauthorised access. Authorization;
Network security;
Security of communications between IoT devices,
Communication and Networking Firmware; [62,72,119–136]
risks associated with open networks.
5G and 6G networks
Machine learning;
Blockchain; [35,59,61–65,69–73,75–
Integrating emerging technologies into IoT
Emergent technologies Artificial intelligence; 93,96,102,110,119–
security solutions.
Edge Computing; 121,123,127–130,132,137–144]
Fog Computing
Identify, address, and mitigate potential risks
Risk management Risk management frameworks [145–152]
associated with security and privacy in IoT.

3.1. Attack Detection

With the spread of IoT devices, cyberattacks favored by the poor security of these
devices have also increased [60]. The attacks can address different levels of the system such
as sensor, network, support, or application. These attacks are intended to cause damage
to the system or to gain unauthorized access to the system or its data [61,66]. The larger
the area the system encompasses, the more damage these attacks can generate. Also, the
rapid spread of the 5G network thanks to the expansion of IoT systems and the increase
in data volume has enabled the development of innovative applications, but at the same
time has also led to an increase in network-level attacks [62,69]. In order to prevent these
attacks, network level intrusion detection systems have been developed that are capable of
detecting anomalies in data transmission between devices [69,70]. IoT networks differ from
traditional ones; thus, it is necessary to develop advanced intrusion detection systems; in
most studies, the use of ML is recommended; however, this poses new challenges [66,73].
First of all, the models need to be trained; the lack of the necessary amount of data and
the disadvantage of the long training duration intervene here [72]. Secondly, there is the
problem of the adaptability of the models to new conditions materialized by new attack
methods [65], detection, and adaptability to concept drift.
One specific type of cyberattack explored in the selected articles is those initiated
by botnet armies. These botnets exploit the vulnerabilities of smart devices connected
to IoT systems, which users often neglect to secure properly [61,77]. Unlike traditional
Internet-connected devices such as computers or smartphones, which typically benefit
from robust security measures, smart home appliances are frequently overlooked, despite
their internet connectivity and inherent risk exposure.
Although botnet detection solutions exist outside of IoT ecosystems, their effective-
ness significantly diminishes within IoT environments [61,63,76–78]. Among the attacks
facilitated or intensified by these botnets, Distributed Denial of Service attacks stand out as
a prominent threat [59,75]. Figure 7 shows typical DDoS attack components using botnets,
managed by a botmaster.
 Although botnet detection solutions exist outside of IoT ecosystems, their effective-
ness significantly diminishes within IoT environments [61,63,76–78]. Among the attacks
facilitated or intensified by these botnets, Distributed Denial of Service attacks stand out
Computers 2025, 14, 61 14 of 45
as a prominent threat [59,75]. Figure 7 shows typical DDoS attack components using bot-
nets, managed by a botmaster.

Botmaster

Bot army

IoT Devices

Figure 7. DDoS attack by botnets.

Figure 7. DDoS attack by botnets.
3.2. Data Management and Protection
3.2. Data Management
This categoryand Protection
includes studies on the management and protection of data in IoT
Thissystems,
category highlighting the vulnerabilities
includes studies that canand
on the management ariseprotection
in data storage
of dataand transfer.
in IoT sys- This is
a critical topic
tems, highlighting the given the huge amount
vulnerabilities that can of dataincollected,
arise data storagestored,
andprocessed and transmitted
transfer. This is a
within
critical topic the system.
given the hugeItamount
also changes
of datathe way data
collected, is accessed.
stored, Before,
processed and users received data
transmitted
within thefrom a specialised
system. It also service.
changes IntheIoT, users
way datacan
is communicate
accessed. Before, directly
userswith sensors.
received dataThey can
obtain dataservice.
from a specialised directly,Inbut they
IoT, cancan
users also transmit instructions
communicate to devices
directly with sensors.[92]. In this
They can context,
obtain data directly, but they can also transmit instructions to devices [92]. In this context,ensuring
there is an increasing need to ensure a seamless data flow, while at the same time
data
there is an privacy through
increasing secure,aefficient,
need to ensure seamlessanddatascalable identity
flow, while management
at the [92].
same time ensuring
data privacy The problem
through of ensuring
secure, efficient,trust
andmanagement, datamanagement
scalable identity confidentiality, and integrity arises in
[92].
Thethe deployment
problem of IoTtrust
of ensuring systems in domainsdata
management, suchconfidentiality,
as the medicaland andintegrity
automotive industries,
arises
as well as inofthe
in the deployment IoTfinancial
systemssector, where critical
in domains such asdecisions
the medicalare made based on the data
and automotive
provided by the system [84]. The application of Blockchain technology [84–86] is becoming
a method of interest due to the security features it can bring by incorporating it into IoT
systems, but there is the issue of scalability and interoperability.
Due to the Internet connectivity of IoT devices that have access to personal data,
problems related to the digital identity of users arise, such as unauthorised access to data
or identity forgery [87,88,91]. There is, thus, a need to develop advanced identity systems
capable of reducing unauthorised access.
There is a trend to use Generative AI within the IoT ecosystem to make it more efficient,
but this integration leads to new vulnerabilities and risks [89,143]. Due to the large volume
of data handled by Generative AI technology, poor system protection can lead to data
privacy breaches and damage data integrity. Also, the generation process itself may contain
risks of information leakage [90].

3.3. Securing Identity Management

Identity security management is a critical component of data protection in IoT en-
vironments, primarily focusing on authenticating entities involved in data transfer and
granting them the necessary authorization [98]. This process is essential to restrict access to
sensitive information exclusively to authorized users and devices, thereby mitigating risks
of unauthorized access and security breaches [99,100].
Device identification involves recognizing and categorizing devices connected to the
network by analysing their distinctive attributes derived from data traffic [73]. Traditional
methods for device identification face challenges in terms of adaptability to newly con-
Computers 2025, 14, 61 15 of 45

nected devices and are prone to errors. Emerging solutions leverage blockchain technology,
but these approaches present limitations, including the potential exposure of sensitive meta-
data, which could compromise user privacy [85], and the challenge of achieving scalability
while maintaining data security [97]. Relying on centralized servers for authentication
introduces vulnerabilities such as a single point of failure [96].
Traditional user authentication methods, such as credentials, certificates, and 2FA,
can pose challenges in an IoT system due to the limited resources and capabilities of the
devices [106,107]. Furthermore, vulnerabilities arising from poor identity management
can be exploited in this context, given the specific characteristics of devices connected to
the system.
Studies [101–103] highlight the importance of secure communication and controlled
access to stored data. Currently, these processes are often managed through PKI [104,105].
While PKI has been an effective standard for securing communication in traditional sys-
tems, emerging challenges in the context of large-scale IoT suggest that it may struggle to
meet increasingly complex requirements. Moreover, there is no well-defined protocol for
efficiently transferring trust or updating PKI credentials when the responsibility for device
maintenance transitions from one service provider to another [104,105].
Key issues include the high costs associated with implementing and maintaining a
PKI system, as well as its substantial resource requirements, which can pose significant
obstacles for organizations aiming to deploy large-scale IoT solutions [104,105]. Another
notable concern is the risk of a single point of failure, where the entire responsibility for
access authorization relies on the PKI infrastructure [101–103]. This means that if the PKI
infrastructure is compromised or becomes inoperative, the entire security framework of
the IoT network could be severely impacted.
The adoption of digital identities introduces additional challenges. For instance,
study [87] draws attention to issues in the medical field related to identity management.
Digital identities have proven insufficient for accurately identifying patients, with a lack of
system integration and limited scalability further complicating the situation. Study [88]
examines the limitations of blockchain-based digital identities in terms of authenticity
and controllability while also addressing privacy requirements. Challenges also arise in
establishing a trusted network and coordinating digital identity management [91].

3.4. Communication and Networking

To develop a massive IoT ecosystem, it is essential to ensure secure communication
and scalable networks that meet security and performance requirements [72] in the con-
text of a large number of devices with limited resources [119–121]. The creation of such
networks requires the implementation of specific protocols, each tailored to the scope of
the IoT system in question. These protocols are fundamental to guarantee the integrity,
confidentiality and availability of data transmitted between devices. They impose security
measures that are essential to cope with cyberattacks, such as DDoS attacks, communication
eavesdropping, or man-in-the-middle attacks [125].
Firmware plays a crucial role in the communication within an IoT ecosystem, as it
directly affects how devices connect, communicate, and interact with the system. Conse-
quently, network-level security must be analysed with consideration of potential firmware
vulnerabilities. Paper [124] highlights this aspect by reviewing studies that focus on ad-
dressing these vulnerabilities.
As previously mentioned, the introduction of 5G networks has brought a series of
vulnerabilities, partly due to its specific features and partly because of device limita-
tions [62]. Its potential successor, the 6G network, is continuously undergoing tests and
studies. The integration of AI technology with 6G in IoT systems offers significant oppor-
Computers 2025, 14, 61 16 of 45

tunities but also introduces new challenges, particularly with the anticipated increase in
the number of connected devices and the volume of data transmitted [123]. This growth
necessitates optimising energy consumption and resource allocation to meet performance
requirements [122].

3.5. Emergent Technologies

The integration of artificial intelligence and machine learning into IoT systems signifi-
cantly enhances cognitive capabilities [78,85,89,90,123,141] and offers a promising approach
for detecting and mitigating cyberattacks in IoT environments [62,76]. The reviewed articles
include studies focused on ML algorithm [35,59,62,65,73,75,137–139]; others exploring deep
learning through the use of neural networks [70,72]; and some examining ML training
techniques such as transfer learning, federated learning, and split learning [71,77,102,140].
However, implementing these technologies introduces new challenges due to the nature
of connected devices, the substantial resource demands required for their deployment,
and the time-intensive process of model training. Additionally, adapting these models to
real-time conditions remains a significant difficulty.
There is also a significant number of studies of blockchain technology in IoT sys-
tems. Blockchain, due to its characteristics of data immutability, decentralisation, and
transparency [84,86–88,91,92,96,102,138], has become a point of interest to secure the IoT
system. Blockchain technology provides a decentralised network; it eliminates the single
point of control; thus, attacks on the system become much more difficult [31]. Challenges
in blockchain arise due to the heterogeneity of devices using different communication pro-
tocols and relying on other technologies and requiring connection to blockchain. Response
delays introduced by transaction confirmation in blockchain can be a negative aspect.
To address the need for processing large volumes of data from diverse devices, edge
computing has emerged as a solution to enhance computational performance in IoT sys-
tems [69,119,121]. This approach involves positioning computational resources closer to the
data source at the network’s edge [123]. Additionally, edge computing can be integrated
with fog computing to facilitate IoT interoperability with Cloud technology [64]. However,
incorporating these technologies into IoT systems introduces new security challenges,
particularly when compromised devices launch attacks targeting fog layer services.

3.6. Risk Management

Risk management in IoT systems plays a critical role in assessing and addressing
cyber risks that could impact the system. Several types of risks can be identified, including
those related to IoT ethics, data security and privacy risks, and technical risks [145]. The
complexity of IoT systems poses significant challenges in analysing and identifying these
risks [147]. Human involvement, the diversity of IoT application domains, and IoT-specific
cybersecurity challenges add further difficulties to the risk management process [146].
Additional challenges include the lack of robust management strategies, the absence of
standardised IoT security measures, and a reactive approach to developing strategies in
response to attacks rather than adopting a proactive stance [148].

4. Identified Challenges and Solutions

4.1. Attack Detection
An essential step in securing the IoT ecosystem lies in the detection of attacks. IoT
network security specialists are focused on developing the most effective methods for
detecting and preventing cyberattacks, aiming to mitigate their impact on critical infras-
tructures as well as on sensitive data. While numerous solutions have been proposed and
analysed to address the current challenges in IoT security, these approaches are not without
Computers 2025, 14, 61 17 of 45

vulnerabilities. What follows is an overview of recent proposals put forth by researchers to

enhance security within IoT systems.

4.1.1. Intrusion and Anomaly Detection and Concept Drift Detection and Adaption
To protect IoT infrastructures, it is essential to employ two major categories of systems:
Intrusion Detection Systems and Intrusion Prevention Systems [60]. The study in [60] fo-
cuses specifically on developing an anomaly detection system, analysing various detection
techniques within IoT ecosystems while identifying several challenges and limitations of
current methods. To address these challenges, the authors propose integrating Incremental
Learning, Transfer Learning, and Deep Learning techniques to develop scalable detection
models capable of continuous updates, enhancing system performance, and reducing
costs and resource requirements. These models can also adapt to contextual changes, a
phenomenon known as concept drift.
Another approach to developing an efficient detection system is presented in [62],
where the authors examine detection methods used in IoT, including signature-based
recognition, anomaly-based detection, hybrid methods, and collaborative approaches
among IoT devices. They also draw comparisons between their strengths and weaknesses.
Collaborative methods are further explored in [66] to ensure information availability during
an attack. This approach relies on secondary devices supporting primary devices in case of
an attack, ensuring the continuity of critical information delivery to users. By employing
redundancy and cooperation among devices, this strategy enhances the system’s resilience
and availability in attack scenarios.
The effectiveness of modern methods based on emerging technologies is also high-
lighted in [62], which discusses the development of detection systems based on Deep
Learning. This technique has proven highly effective in detecting attacks within 5G net-
works. Using deep learning, intrusion detection pipelines have been created to leverage
powerful algorithms capable of identifying and mitigating security threats in real time [69].
An adaptive and high-performing IDS was implemented in the context of electric
vehicle charging stations using neural network architectures that combined LSTM and
GRU models [70].
One challenge in implementing IDS systems is the prevalence of false-positive alarms.
To address this issue and improve classification accuracy, TL and the CBAM [71] can be
used. These techniques, through the utilization of channel and spatial attention, refine
feature maps for greater precision.
In anomaly detection systems, careful consideration must be given to the selection of
the network architecture, as it is a key factor in achieving more effective anomaly detection.
This was demonstrated in [72,74] where two architectures, EPA and MUD, were compared.
The authors showed the superior performance of EPA over MUD. While MUD focuses
solely on stateless communication states, EPA provides a comprehensive evaluation of all
communication states, offering more detailed analysis for anomaly detection.
Further advancements are presented in a Deep Learning-based IDS for IoT devices, ca-
pable of detecting diverse attack types, including Blackhole, DDoS, Sinkhole, and Wormhole
attacks. The system employs a four-layer deep Fully Connected (FC) network architecture,
making it communication protocol-independent and reducing deployment complexity [83].

4.1.2. DDoS Attacks

The reviewed articles highlight a strong interest in addressing specific cyberattacks
that can cause significant damage. For instance, study [64] proposes a tailored solution
for detecting DDoS attacks, considering the phenomenon of concept drift. The solution
involves an adaptive online framework capable of adjusting its performance in real-time
Computers 2025, 14, 61 18 of 45

based on changes in the network environment. Concept drift detection is achieved using
ADWIN and DDM methods, while learning capabilities are enhanced through ARF, SRPs,
and KNN methods.
One significant challenge in developing an effective framework for detecting DDoS
and Botnet attacks is the imbalanced and limited availability nature of data for accurately
simulating such attacks. The scarcity of comprehensive real-world datasets constrains
the ability to train robust and generalizable detection models. Moreover, many tradi-
tional detection solutions rely on unlabeled or untrustworthy datasets, which can degrade
model performance, particularly when faced with zero-day threats [75]. To address this
limitation, studies [59,75] proposed leveraging Conditional Tabular Generative Adversar-
ial Networks (CTGAN) to generate synthetic data that closely mimics real-world traffic
patterns. This approach not only enriches the training datasets but also incorporates a
discriminator framework, which enhances the system’s capability to accurately distinguish
between legitimate and malicious traffic, thereby improving the overall effectiveness of the
detection mechanism.
To address the class imbalance issue, researchers [79] have explored the use of en-
semble learning techniques, such as the Bagging classifier, which employs a deep neural
network as a base estimator. By incorporating class weights into the training process, this
method ensures the creation of balanced training subsets for the DNN, improving both the
coherence and effectiveness of intrusion detection and classification systems.
In [35], a solution is proposed for detecting DDoS attacks in Information-Centric
Networking for IoT networks using machine learning algorithms such as SVM, RF, and
KNN. However, the best results were obtained by applying DT and RF classifiers [65]
trained on features selected using GA.
Feature extraction was further improved by converting non-image data into image data
through deep learning techniques, particularly VGG16 and Inception [71,72]. The Inception
technique, specifically the TCN model within the Inception structure, is proposed in [73] for
identifying devices connecting to the network. This method focuses on packet feature extraction,
feature selection, and, ultimately, extracting the temporal characteristics of the packets.
To address the challenges associated with IoT devices’ limited computational re-
sources and storage capacities, a lightweight and efficient intrusion detection method has
been proposed [82]. This solution incorporates a fast protocol parsing approach on raw
packet capture files to generate semantic-level features, followed by session merging and
feature grouping techniques to improve detection accuracy. These characteristics make
it an efficient, extensible, and suitable approach for IoT intrusion detection in resource-
constrained environments.

4.1.3. Botnet
As IoT systems proliferate, the risk of botnet-driven attacks also increases. The study
in [63] examines traditional attack detection methods, which, despite their high resource con-
sumption, are effective in identifying attacks generated by IoT-based botnets. Such approaches
can serve as a valid starting point for developing new detection and prevention techniques.
To address the limitations of traditional methods in the IoT context, a botnet at-
tack mitigation framework called IMTIBot was developed [61]. This framework seg-
regates network traffic into normal and abnormal categories and leverages ensemble
learning classifiers, combining multiple machine learning models to enhance detection
accuracy. Another innovative solution is the strategic amalgamation of Hybrid Feature
Selection methods—Categorical Analysis, Mutual Information, and Principal Component
Analysis—with an ensemble of machine learning techniques [81]. This approach refines
the input space for ensemble learners, with Extra Trees as the primary technique.
Computers 2025, 14, 61 19 of 45

Paper [76] introduces BotStop, a machine learning-based framework for detecting

botnet activity in IoT devices through the analysis of individual network packet features.
The approach emphasizes the selection of a minimal set of seven essential features.
Another botnet attack detection framework is proposed by [78]. This study presents a
lightweight deep learning approach for detecting five types of botnet attacks—DoS, DDoS,
fuzzing, Boofuzz, OS fingerprinting, and port scanning—in IoT networks. The proposed
model, designed with a streamlined architecture featuring four convolutional layers and
global average pooling, achieves high classification performance with minimal computa-
tional and memory requirements. The approach eliminates the need for extensive feature
engineering, providing an efficient and scalable solution for real-time botnet detection.
A novel optimisation-based solution addresses the persistent challenge of low detec-
tion accuracy in IoT botnet detection [80]. By improving the initial population generation
strategy of the Dung Beetle Optimiser (DBO) with a centroid opposition-based learning
approach, this method optimises Catboost parameters for enhanced detection performance.
Article [77] proposes a decentralized model for mitigating DDoS attacks in corporate
local networks by integrating Host Intrusion Detection Systems (HIDS) and Network
Intrusion Detection Systems (NIDS) with federated learning. Deployed within a fog
computing infrastructure, the model enables real-time detection and mitigation of malicious
traffic while preserving privacy and reducing the risk of a single point of failure.

4.1.4. Eavesdropping Attacks

The studies [66,67] address the issue of eavesdropping, a challenge that has received
relatively little attention in specialised literature. The collaborative method described in [67]
ensures signal accuracy for devices within the network while simultaneously disrupting
signals to devices attempting unauthorised interception of messages.
In [67], a BP neural network model is proposed for detecting eavesdropping attacks in
environments with a low signal-to-noise ratio. Meanwhile, study [68] highlights infrared
communication and the risk of “listening” to signals emitted by remote controls. To prevent
data theft in this context, the authors propose an encryption method that regenerates keys
each time the remote control’s power button is pressed.
Table 4 provides an overview of the key challenges and solutions in attack detection,
emphasizing the main issues and proposed strategies to mitigate them.

Table 4. Key challenges and solutions in attack detection.

Challenge Related Challenges Key Threats Solutions

Integration of ML techniques such as Incremental
Anomaly detection Managing data diversity and Limited scalability and resilience in
Learning, Transfer Learning, and Deep Learning to obtain
in IoT scalability in the IoT ecosystem detecting cyberattacks
scalable and adaptable models able to handle concept drift
References [60,69–71,83]
Detection and Response time optimisation, Continuous evolution of DDoS, Botnet
Using ML techniques to improve response time, system
Prevention of DDoS and limited computational resources attacks, and inability of the system to
adaptability, and network traffic classification
Botnet attacks of devices adapt in real time
References [35,59,61–65,72,73,75–78,80–82]
High number of false alarms, Use of ML methods for intrusion detection, collaborative
Anomaly detection High resource consumption required
balancing detection accuracy and systems for effort sharing;
efficiency by traditional detection systems
resource consumption Selection of the right architecture
References [61,66,71,74,79]
Introducing intentional signal perturbations to
Balancing the effectiveness of signal disrupt eavesdroppers;
Unauthorised interception of
disruption for malicious devices Backpropagation neural network model specifically
Eavesdropping attack communication signal, difficulty
without degradation of quality for designed for detecting eavesdropping attacks in low
detection of detection in low signal-to-noise
legitimate users, detection of SNR scenarios;
ratio environments
interception when signal is weak Signal encryption or modulation techniques to protect
against unauthorised interception
References [66–68]
Computers 2025, 14, 61 20 of 45

4.2. Data Management and Protection

In the domain of data management and protection, the primary challenges revolve
around device and user authorization, ensuring data integrity, and maintaining data
confidentiality. Effective solutions must address the verification of identities to prevent
unauthorised access, protect data from unauthorised modifications to guarantee its accu-
racy, and implement robust encryption mechanisms to safeguard sensitive information
from breaches and interception.

4.2.1. Data Security and Privacy

Building on the Hyperledger Fabric framework, the authors of [84] propose an in-
novative concept based on the idea of Blockchain as a Service (BaaS) for securing and
protecting data. This integration is achieved through a novel architecture combined with an
encrypted data structure utilising public and private keys, offering a high level of security
for data management.
Conversely, study [92] introduces a blockchain-based platform that leverages smart
contracts to enhance data protection. This solution builds upon a three-tier architecture
with the addition of a new layer called the Blockchain Composite Layer. This extra layer
improves functionality and security, enabling decentralised and automated management of
transactions and data. To increase trust in transactions within Ethereum-based blockchain
frameworks, study [85] proposes introducing a legitimacy rating mechanism through a
consensus method and a decentralised proof matrix. Cloud environment security is further
enhanced using neural networks for anomaly prediction, providing an additional layer of
protection against emerging threats.
Expanding on these approaches, study [93] proposes a lightweight group management
model for IoT networks using Hyperledger Fabric, aimed at improving data security. It
utilizes group keys to ensure that only authorized users within a group can access sensitive
data, thereby reducing the risk of information leakage. To address the overhead of rekeying
in resource-constrained IoT devices, the model introduces a trusted agent for efficient key
distribution. This approach enhances network lifetime, reduces storage costs, and improves
processing time compared to existing methods.
The study in [86] emphasises cloud data security, proposing blockchain technology
combined with a distributed agent model as a solution. Files are assigned a unique hash
value generated using a Merkle hash tree, enabling continuous monitoring to verify their
integrity. In case of discrepancies, real-time alerts are sent to the file owners.
Extending these approaches, study [94] introduces an innovative integration of
lightweight blockchain technology within IoT systems designed to mitigate the computa-
tional overhead typically associated with conventional blockchain implementations. This
integration not only streamlines the implementation process but also reduces overall com-
plexity. Furthermore, the incorporation of the Okamoto–Uchiyama encryption algorithm
significantly bolsters data privacy. As a result, the proposed framework establishes a secure,
decentralized platform for the storage and analysis of sensitive supply chain data, allowing
decentralized applications to perform computations on encrypted data while ensuring
data confidentiality.
Study [95] presents a multi-dimensional chaotic encryption scheme to enhance data
security in IoT systems. By leveraging fixed-point operations, it minimises computa-
tional overhead and power consumption, making it suitable for resource-constrained
environments. A chaotic dynamic analysis scheme improves system evaluation, while
a multi-dimensional encryption method enhances sequence randomness, strengthening
cryptographic resilience. This adaptable framework enables optimised security configura-
Computers 2025, 14, 61 21 of 45

tions based on data sensitivity and real-time demands, reinforcing privacy and secure data
transmission in decentralized IoT networks.

4.2.2. Digital Identity and Identity-Based Encryption

The study in [86] highlights the role of blockchain technology in securing digital
identity through decentralised identity solutions, consent management, and lifecycle man-
agement to ensure relevance and accuracy. Blockchain technology also addresses challenges
such as scalability and unauthorised access.
In blockchain-based digital identity systems, proposed solutions for enhancing data
security include separating identity verification from credential issuance, utilising linkable
ring signatures to protect the verifier’s identity, employing cryptographic methods for
revocation to maintain privacy, and leveraging smart contracts for system management
and auditability [88].
Another blockchain-based solution is proposed in [91], featuring high-resistance dy-
namic encryption, encrypted SSL-VPN channels, and dynamic key mechanisms. The
proposed system emphasises anonymous authentication, robust security classifications,
and access controls to prevent unauthorised data access and brute-force attacks.

4.2.3. Generative AI
Protecting data privacy and integrity in the context of the proliferation of Gen-
erative AI is crucial. In this regard, the authors of [89,90] propose a multi-faceted
approach that includes techniques such as encryption, anonymization, access control,
continuous monitoring, protocol development, multi-layered security mechanisms, and
AI-powered safeguards.
Federated Learning combined with partial training can protect privacy in machine
learning applications within IoT systems [90]. In this approach, IoT devices train smaller
sub-models based on a large model hosted on a cloud server, and the server aggregates
these sub-models to update the global model. TEE are employed to secure sensitive user
data, protecting it from external threats before it is sent to generative models for inference.
Table 5 summarises the key challenges and solutions in data management and protec-
tion, highlighting the main issues and proposed strategies to address them.

Table 5. Key challenges and solutions in data management and protection.

Challenge Related Challenges Key Threats Solutions

Blockchain-based frameworks
(Hyperledger Fabric), decentralised
Ensuring data integrity and secure
Data access by unauthorised entities data management, encrypted data
Data Privacy and Security storage on decentralised networks and
and attacks on data integrity structures, and federated learning to
in the Cloud
ensure data privacy by preventing
unauthorised access
References [84–86,92–95]
Handling a large number of Separation of identity verification and
transactions and identity verifications credential issuance;
Mitigating unauthorised access and efficiently in a decentralised system, Linkable ring signatures, smart
Securing Digital Identity maintaining accurate lifecycle protecting against brute-force and contracts, encrypted
management of identities advanced cryptographic attacks, SSL-VPN channels;
ensuring encryption mechanisms are Robust security classifications and
robust and dynamic access controls
References [87,88,91]
Protecting sensitive data across Employing encryption, anonymization,
Data breaches, unauthorised access,
distributed systems while balancing and multi-level security mechanisms,
Data privacy and integrity in context of exploitation of sensitive user inputs,
security, computational efficiency, and Using Trusted Execution Environments
Generative AI technologies and privacy leakage during Federated
privacy during AI model training, to protect data inputs during
Learning model aggregation
aggregation, and inference model inference
References [89,90]
Computers 2025, 14, 61 22 of 45

4.3. Securing Identity Management

Identity security management in IoT addresses key issues such as authentication,
authorization and identity management of connected devices. Researchers are exploring
and developing new protocols, technologies, and frameworks to ensure secure interactions
in the IoT ecosystem, given the limited resources and vulnerability of devices.

4.3.1. Device Identification

Paper [73] proposes a device identification scheme based on extracting time series
characteristics of data packets, which are subsequently used as unique fingerprints of the
devices. Another approach, presented in [101], involves using the wireless channel state
characteristics of devices for identification. Although wireless channels can be unstable,
this drawback is compensated for by using a locally sensitive hashing algorithm, which
improves the stability and accuracy of the identification. An alternative method, based
on Paillier homomorphic encryption, is described in [100], allowing verification of device
identity without decrypting the message, an efficient approach for privacy preserving.
Adding to these advancements, study [108] introduces a mutual authentication proto-
col that integrates Physically Unclonable Functions (PUFs) as a hardware-based security
measure. This approach replaces static secret keys with dynamic responses derived from
the physical characteristics of devices, ensuring enhanced resistance to physical and cloning
attacks. To address the inherent noise in PUF outputs, a Fuzzy Extractor (FE) is employed
for consistent cryptographic key generation. Further, study [109] refines the use of PUFs
by proposing a lightweight authentication scheme that leverages geometric threshold
secret-sharing to avoid explicit storage of challenge-response pairs. This design mitigates
risks such as side-channel and machine-learning attacks while maintaining computational
efficiency, making it well-suited for resource-constrained IoT environments. Study [112]
proposed a Firmware-Secure Multi-Factor Authentication to enhance both the physical
and software security of IoT devices. FSMFA integrates PUFs with firmware integrity
verification to enable mutual authentication and secure key negotiation between devices
and servers. Additionally, it incorporates a challenge-response mechanism and a secure
firmware update scheme to ensure security throughout the device lifecycle.
To address vulnerabilities in traditional PUF-based schemes, a quantum-safe authenti-
cation method is proposed [113], utilizing CRYSTALS-Kyber homomorphic encryption and
a two-server model to secure PUF responses without helper data. Complementing these
methods, study [118] underscores the critical IoT security challenges—such as weak pass-
word enforcement, unencrypted communications, and physical sensor tampering—and
reinforces the benefits of a PUF and lightweight encryption technique-based solution for
energy-constrained devices.
Expanding on these approaches, study [110] introduces a blockchain-based IoT frame-
work that integrates advanced computational methods for secure device identification
and authentication in digital healthcare systems. By utilizing a hybrid predictive model
combining CNNs and GRUs, this system extracts complex patterns and critical directional
features from IoT-generated data. Coupled with the Jellyfish Search Optimisation (JSO)
algorithm for feature selection and the Twofish encryption algorithm for data protection,
the framework ensures both robust device identification and secure data management.
On the other hand, the paper in [102] proposes a multi-layered solution for securely
distributing data to users in IoT networks based on blockchain technology. This solution
manages authentication, key, and message exchange in a decentralised and secure way.
The framework uses the ACE protocol for data encryption, ensuring robust protection of
information transmitted between devices and users.
Computers 2025, 14, 61 23 of 45

To prevent attacks and ensure the authenticity and integrity of data, study [107]
proposes a framework based on Bloom filters and hash chains. This system could serve
as a viable solution in the context of an increasingly complex IoT ecosystem, providing
enhanced protection against cyberattacks and ensuring a secure and efficient data flow.
Digital identity, as a method of representing devices within the IoT ecosystem, is dis-
cussed in works [87,88,91,96]. Despite the advantages of blockchain-based digital identity
systems, they have several drawbacks, such as issues with identity authenticity, controllabil-
ity, and privacy protection [88,96]. The study in [87] provides an overview of the challenges
and solutions in the medical field. To address these weaknesses, works [88,96] propose
a system where the roles of identity verification and credential issuance are separated to
reduce the risk of identity-related information leakage. Privacy is enhanced by linkable
ring signatures, zero-knowledge proof encryption techniques, and AES. Using a similar
approach, the authors of [88,96] developed a Multi-Factor Authentication method utilising
blockchain and zero-knowledge proofs. They address weaknesses such as single points of
failure and privacy vulnerabilities in blockchain technology through a DAM. Part of the
proposed MFA process also includes using NFTs as authentication tokens.
The paper at [91] identifies several dimensions of digital identity characteristics for
users. It proposes a collaborative framework between governmental institutions and non-
governmental blockchain alliances, based on a delegated model. It proposes a zero-trust
model for digital identity management and big data security.

4.3.2. Authentication
A new framework proposed in [97], based on edge computing and blockchain, explores
the use of Ethereum 2 Layer roll-ups to enhance scalability and reduce bottlenecks in the
device authentication process. This approach could alleviate the pressure on authentication
systems and enable more efficient resource management, given the exponential growth in
the number of connected devices. At the same time, study [85] introduces an Ethereum-
based mechanism that ensures data security through a unique legitimacy score, applicable
both at the device level and at the cloud level.
Regarding authentication, study [106] suggests a mutual authentication and key
agreement protocol designed to address threats in the edge–fog–cloud architecture of 5G
networks. This protocol involves mutual identity verification between devices and fog
nodes, adding an extra layer of security in the resource access process.
A significant enhancement in handover authentication protocols is proposed in [111],
which addresses critical deficiencies in traditional methods. Traditional handover au-
thentication protocols, often reliant on bilinear pairing and elliptic curve cryptography,
are susceptible to quantum attacks and session key compromise. To overcome these vul-
nerabilities, refy introduces a lightweight two-party handover authentication protocol
based on the lattice cipher NTRU, designed to resist quantum attacks. This protocol elimi-
nates the dependency on a home agent, reducing communication delays and improving
session key security. Expanding on lattice-based cryptography, ACPRE enhances proxy re-
encryption by embedding dual access policies and securing data via the LWE problem [115].
It achieves HRA security with formal proofs while optimising efficiency through plain-
text space expansion, reducing performance overhead and complementing lattice-based
handover authentication.
In addition to advancements in handover authentication, securing digital data ex-
change in IoT environments against quantum threats is paramount. Traditional crypto-
graphic schemes struggle against the computational power of quantum computers, necessi-
tating novel encryption frameworks. A recently proposed approach integrates bit-plane
extraction, chaotic sine models, hyperchaotic maps, and quantum operations to enhance
Computers 2025, 14, 61 24 of 45

data security [114]. By leveraging quantum-state superposition, chaotic diffusion, and

selective scrambling at the bit level, this method ensures robust encryption while maintain-
ing efficiency for real-time IoT applications. Experimental results confirm its effectiveness,
demonstrating strong security metrics and rapid execution.
A flexible and secure IoT access control scheme enhances user identity sovereignty
while mitigating single points of failure. Unlike traditional CP-ABE schemes, it decen-
tralizes key generation via proxy clusters and employs self-sovereign identity for privacy-
preserving attribute validation [116]. A modified CP-ABE scheme is proposed [117] to
enhance data confidentiality and access control in IoT-based healthcare by offloading
computationally intensive encryption tasks to multiple cooperative nodes. This approach
optimises workload distribution based on node capacity, reducing computation time and
energy consumption.
Another notable contribution, study [103] proposes a one-time pad protocol to ensure
secure communication in IoT, where keys are generated through a multiparty sum of
random numbers derived from noise and physical phenomena detected by sensors. This
method adds an additional layer of security by using natural phenomena in the encryption
process. Similarly, study [98] explores the use of sensors to support authentication, sug-
gesting that factors such as the sensor’s state and the environment in which it operates can
play a crucial role in determining a device’s legitimacy. Furthermore, study [99] proposes
a three-phase authentication protocol, the first phase being user registration, followed by
data encryption using the ECC-AES model and key generation via the SI-AO.
The study [104] addresses the lack of protocols for trust transfer from one service
provider to another by developing a framework that minimises the need for manual
intervention by automating the IoT device registration process and issuing operational
certificates for new service providers. Paper [105] introduces a new architecture to eliminate
the single point of failure issue in the use of PKI, which is easily applicable in IoT systems
with resource-constrained devices. This architecture involves the use of ECC cryptography,
certificates, and a decentralized PKI system divided into zones, with each zone having a
master zone responsible for the devices within that specific zone.
Table 6 provides a summary of the key challenges and solutions in identity security,
outlining the main issues and proposed approaches to address them.

Table 6. Key challenges and solutions in securing identity management.

Challenge Related Challenges Key Threats Solutions

Blockchain and Edge Computing based
multilevel frameworks;
Unauthorised access, data breaches, Cryptographic techniques like zero-knowledge proofs,
instability of wireless channels, single AES, ring signatures, distributed authentication
points of failure, identity privacy mechanisms, and secure data-sharing protocols;
Device identification Device identification management
vulnerabilities, and insufficient Device identification using time series;
protection in IoT and blockchain-based Physically Unclonable Functions with Fuzzy Extractors;
identity systems Geometric threshold secret-sharing in PUFs;
Firmware-Secure Multi-Factor Authentication;
Zero-trust digital identity model;
References [73,87,88,91,96,100–102,107–110,112,113,118]
Using Ethereum Layer 2 roll-ups; mutual authentication,
decentralised PKI, one-time pad encryption,
sensor-based verification, ECC-AES encryption, and
Securing credentials in Increased vulnerability due to limited automated IoT trust transfer;
Authentication
low-resource environments resources in the authentication context Lattice cipher NTRU based protocol;
Lattice-based proxy re-encryption (ACPRE) with dual
access policies;
Quantum-enhanced encryption frameworks;
References [85,97–99,103–106,111,114–117]
Computers 2025, 14, 61 25 of 45

4.4. Communication and Networking

Communication and networking in IoT relies on different protocols and technologies,
i.e., a variety of networks, each of which involves certain vulnerabilities. Some of the
studied articles also propose solutions to mitigate these risks.

4.4.1. Network Security

The solution proposed in article [72], described in the Attack Detection subsection, also
addresses the issue of network security and communication between system components
by improving device identification. This ensures enhanced network security against device-
specific attacks. For detecting representative attacks within a network, study [120] proposes
a solution utilising 110 neural networks. Additionally, it improves the attack-sharing loss
function, reducing the number of false alarms and thus contributing to a higher detection
rate of actual attacks on the system.
In study [125], the focus began with the goal of increasing the security of the MQTT
protocol, ideal for use in systems where devices have limited resources. To this end, it
concentrated on the impact of task-specific feature selection. For anomaly detection, five
ML algorithms were analysed: DT, KNN, RF, AdaBoost, and XGBoost, with RF proving to
deliver the best results.
A broader comparative analysis in study [128] explored the effectiveness of feature
reduction and ML techniques across datasets, employing six ML models (DFF, CNN, RNN,
DT, Logistic Regression, and Naive Bayes) and three feature extraction algorithms (PCA,
AE, and LDA). While PCA and AE showed strong performance when dimensionality was
optimised, LDA degraded outcomes on certain datasets. These findings underscore the
need for a universal benchmark feature set to standardize NIDS evaluations.
Deep learning techniques have also been extensively utilized for robust detection
systems. Study [126] proposed a NIDS tailored for cloud environments, using a transformer
model with advanced attention mechanisms to analyse feature relationships, enhancing
detection accuracy and adaptability to evolving threats. Similarly, paper [127] addressed
DDoS attacks with an online SDN defense system that combines CNN and LSTM models
for anomaly detection and flow-rule-based mitigation. By tracing malicious traffic back
to its source through IP tracing, this system provides a robust real-time defense against
such attacks.
In a novel approach to large-scale network security threats, a knowledge graph-based
detection method was constructed by combining a feature template with CNN, BiLSTM,
and CRF layers, forming the FT-CNN-BiLSTM-CRF model [129]. This method excelled in
detecting multi-step network attacks, outperforming other techniques in terms of speed
and accuracy.
Article [121] introduces the concept of an IoT Proxy, aimed at offloading security
aspects to a more powerful gateway supplied with VNSFs. This approach would mitigate
the limitations of devices, such as constrained computational capacity and memory. Ad-
dressing the challenges encountered in IoT systems, study [119] proposes a protocol and
an algorithm for grouping devices based on coverage, storage capacity, and power. This
solution would lead to better network scalability, optimised consumption, and improved
load balancing. To enhance IoT network efficiency, study [136] introduces a two-layer
NOMA-based architecture with caching, addressing bandwidth constraints, latency, and
congestion in large-scale deployments in the context of smart cities. By optimising re-
source allocation through block coordinate descent and inner approximation, the approach
maximizes data rates while maintaining low computational complexity.
Computers 2025, 14, 61 26 of 45

4.4.2. Firmware
Another critical aspect in IoT systems is the vulnerability of IoT devices at the firmware
level. Study [134] highlights significant security vulnerabilities in smart home IoT firmware,
revealing ten critical network-based flaws, with five scoring a maximum CVSS of 10.0. The
findings underscore the widespread use of unsafe functions and the absence of essential
security features. Study [124] provides a review of firmware vulnerabilities, identifying
the challenges encountered at this level and methods to mitigate them. To achieve the
desired level of security, the proposed solutions include the development of standards
and guidelines for stakeholders involved in IoT system development, the application of
emerging technologies to deliver intelligent and adaptive solutions, the use of reverse
engineering for firmware analysis, and the development of hybrid frameworks to unify
various approaches.
A key issue in firmware security is ensuring timely and cost-effective updates. To
address this, study [130] proposes a decentralized, blockchain-based firmware update
mechanism. This approach stimulates distributors via smart contracts and rewards IoT de-
vices for successful installations, using verifiable proof-of-delivery and proof-of-installation
to ensure security and fairness.
Another approach focuses on improving Firmware Update Over the Air efficiency,
particularly for IoT devices using LoRaWAN [131]. A proposed modular firmware devel-
opment framework allows partial and dynamic updates without requiring a system reboot,
significantly reducing update size and network traffic compared to traditional monolithic
firmware updates.
Beyond update mechanisms, outdated open-source components in firmware pose
additional security risks, as they often contain unpatched N-day vulnerabilities. Study [132]
introduces VERI, a system for large-scale vulnerability detection through lightweight
version identification. VERI leverages symbolic execution with static analysis to accurately
determine open-source components versions and employs deep learning to extract version-
vulnerability relationships from vulnerability descriptions.
Beyond updates, IoT repackaging presents a serious threat, where attackers modify
legitimate firmware by injecting malicious code before redistribution. To mitigate this,
study [133] introduces PARIOT, a self-protecting scheme that integrates anti-tampering
controls directly into firmware, enabling runtime detection of unauthorized modifications
without relying on internet access, secure storage, or external trust anchors.

4.4.3. 5G and 6G Networks

The challenges introduced by the characteristics of 5G and 6G networks have been
explored in a series of articles. Among the challenges addressed in study [122] are spectrum
scarcity and network security. It proposes solutions such as dynamic spectrum sharing
and blockchain-based security. Study [135] presents a mechanism for dynamic spectrum
sharing, which introduces superior spectral and energy efficiency. This framework is based
on ACEDA algorithm for spectrum allocation decisions. Studies [62,123] complement these
by recommending the use of emerging technologies like AI for anomaly detection at the
network level, aiming to reduce response time and optimise resource consumption [123].
The development of ML methods provides scalability for protection systems as the
attack surface expands, while maintaining efficiency and detection accuracy [62]. Another
proposal from the authors of study [62] focuses on designing and evaluating robust models
based on open, standardised datasets tailored for IoT in 5G/6G environments, which also
incorporate new forms of attacks.
Table 7 summarises the key challenges and corresponding solutions in communication
and networking, highlighting the main issues and proposed approaches to address them.
 Computers 2025, 14, 61 27 of 45

Computers 2025, 14, x FOR PEER REVIEW 28 of 47

Table 7. Key challenges and solutions in communication and networking.

Challenge Related Challenges Key Threats Solutions

Self-protecting anti-tampering firmware
Developing IoT security standards, leveraging emerging
scheme
technologies for adaptive solutions, employing reverse
References Firmware vulnerabilities leading to
[124,130–134] engineering for firmware analysis, and implementing
Ensuring firmware security in
unauthorised access, data breaches, hybrid frameworks for unified security approaches;
Firmware security context of diverse IoT
devicewith
ecosystems
and exploitation by attackers through Grouping devices based
Blockchain-based on capacity
decentralized firmware and
Dealing the diversity of Scalability with increasing
unpatched or outdated software. de- update mechanism;
coverage;
Large-scale vulnerability detection system;
Network Scalability and connected device types and vices connected to the system,
Load balancing
Self-protecting optimisation
anti-tampering protocols;
firmware scheme
Load Balancing resource requirements; impacting load management
References [124,130–134] Dynamic feature selection for efficient data
Optimise resource allocation and resource utilisation
Dealing with the diversity of Scalability with increasing devices processing
Grouping devices based on capacity and coverage;
Network Scalability and connected device types and connected to the system, impacting
References
Load Balancing resource requirements;
[72,119–121,125–129,136]
load management and
Load balancing optimisation protocols;
Dynamic feature selection for efficient data processing
Optimise resource allocation resource utilisation Dynamic spectrum-sharing, AI and block-
Managing high-speed data Spectrum availability and se-
References [72,119–121,125–129,136]chain integration for secure 6G applications
Integrating 6G in IoT transfer, spectrum allocation, curity issues in 6G applica-
Managing high-speed data and protocol
Dynamic development
spectrum-sharing, for real-time
AI and blockchain re-
integration
Integrating 6G in IoT andspectrum
transfer, latencyallocation,
requirements
and tions
Spectrum availability and security
forsponse
secure 6Gin applications and protocol development
issues in 6G applications 6G networks in IoT systems for
latency requirements real-time response in 6G networks in IoT systems
References [72,122,123,135]
References [72,122,123,135]

4.5. Emergent Technologies

4.5. Emergent Technologies
Recent studies are increasingly focusing on developing solutions using emerging
Recent studies are increasingly focusing on developing solutions using emerging
technologies such as machine learning, artificial intelligence, edge computing, behav-
technologies such as machine learning, artificial intelligence, edge computing, behavioural
ioural analytics, and blockchain technology [65]. According to the analysis conducted by
analytics, and blockchain technology [65]. According to the analysis conducted by the
the authors of study [138], there has been a noticeable rise in interest among researchers
authors of study [138], there has been a noticeable rise in interest among researchers since
since 2023 regarding the integration of these technologies into IoT system security meth-
2023 regarding the integration of these technologies into IoT system security methods. To
ods. To provide a comprehensive understanding of the growing interest in emergent tech-
provide a comprehensive understanding of the growing interest in emergent technologies
nologies and their application, it is important to contextualize their integration not only
and their application, it is important to contextualize their integration not only within the
within the specific domain of IoT security but also in broader, interdisciplinary fields. This
specific domain of IoT security but also in broader, interdisciplinary fields. This approach
approach allows for a holistic assessment of the maturity and adoption trajectory of these
allows for a holistic assessment of the maturity and adoption trajectory of these technologies,
technologies, as well as the scale of their potential impact.
as well as the scale of their potential impact.
To support this analysis, a chart, Figure 8, illustrating the overall growth of interest
To support this analysis, a chart, Figure 8, illustrating the overall growth of interest in
in emergent technologies across all research fields is presented. The chart is based on data
emergent technologies across all research fields is presented. The chart is based on data
collected through a comprehensive search of articles in four major academic and publish-
collected through a comprehensive search of articles in four major academic and publishing
ing platforms: MDPI, Science Direct, Springer, and IEEEXplore.
platforms: MDPI, Science Direct, Springer, and IEEEXplore.

Figure 8. Emergent technologies: research interest trends.

Figure 8. Emergent technologies: research interest trends.
The chart results reveal key insights into the research momentum surrounding emer-
The
gent chart resultsSpecifically,
technologies. reveal key the
insights into the research
data highlight a periodmomentum surrounding
of stagnation in research emer-
activity
gent technologies.
until Specifically,
2018, followed by a phasethe of
data highlight
linear growth a period
betweenof 2020
stagnation in research
and 2022. activ-
Notably, from
ity until 2018, followed by a phase of linear growth between 2020 and 2022. Notably, from
2023 onward, there is a marked rise in research interest, signaling a significant shift in
focus toward these technologies.
Computers 2025, 14, 61 28 of 45

2023 onward, there is a marked rise in research interest, signaling a significant shift in focus
toward these technologies.
This trend reflects the increasing integration of emergent technologies across diverse
disciplines, providing evidence that the heightened interest in their application to IoT
security is part of a broader, global research movement rather than a localized phenomenon.
Such widespread growth validates the assertion that these technologies are gaining traction
and aligns with the notion of a global technological shift.
Furthermore, the observed trend indicates a maturing phase for emergent technologies,
as their growing adoption across fields demonstrates progress in their development and
readiness for implementation. This overall increase in research activity suggests that these
technologies are advancing toward higher feasibility and reliability, making them more
suitable for integration into specialized domains, including IoT security.
With this broader context established, the focus shifts to IoT security to examine the
specific contributions of these technologies within the field. The following section explores
key emergent technologies individually, detailing their distinct roles and capabilities in
strengthening IoT system security.

4.5.1. Machine Learning

The use of ML methods is proposed due to their ability to mitigate and prevent cyberat-
tacks by continuously updating databases with potential attack signatures and performing
real-time network traffic analysis for anomaly detection [35,62]. ML proves valuable in
predicting potential threats, making decisions, and optimising resource allocation during
an attack [138]. The reviewed articles include evaluations of algorithms to determine which
is most effective for classification and feature selection [61,65,76,80–82,120,127–129,137,139].
By optimising model complexity and selecting lightweight algorithms, a balance can be
achieved between anomaly detection efficiency and computational performance. However,
one of the significant challenges in developing an efficient ML-based framework lies in
the imbalanced nature of the available datasets, which often feature a disproportionately
low number of malicious instances [79]. This imbalance can lead to imbalanced learning
of the model, where the algorithm becomes biased towards the majority class, thereby
compromising its ability to accurately detect and classify minority class instances such as
rare or novel cyberattacks [59,75].
The increasing adoption of smart home systems, driven by the advancement of IoT
technologies, has amplified the demand for robust security mechanisms to address vul-
nerabilities and ensure user privacy [142]. Traditionally, ML models are deployed on
cloud-based infrastructures with high computational capacity, but this approach introduces
latency issues and exposes user data to privacy risks. In response, on-device ML models
are gaining attention as they allow data to remain local, enhancing security and supporting
real-time applications such as intrusion detection [142].
The use of Deep Learning, as a subset of ML, is recommended due to its capabilities
in processing complex patterns, ensuring efficient detection of anomalies and intrusions,
reducing false positives, and enabling device identification based on unique features
extracted from network traffic [73,83,132]. In this context, advanced neural network models
are analysed, focusing on their architecture, training methodologies, and ability to capture
intricate patterns in high-dimensional data [110]. These models are evaluated for their
potential to enhance predictive accuracy, robustness against adversarial scenarios, and
adaptability to dynamic environments, particularly in scenarios involving complex anomaly
detection and intrusion prevention systems [70,72,78,120]. Additionally, the deployment of
lightweight ML algorithms, such as Decision Trees, has demonstrated superior performance
Computers 2025, 14, 61 29 of 45

in terms of computational efficiency and energy consumption during both training and
inference phases, making them suitable for resource-constrained IoT devices [142].
The reviewed articles also address the issue of selecting a training strategy for ML
models. Transfer learning emerges as a solution to reduce the time and computational effort
required for training new models [71], leveraging prior knowledge during the training
process. Collaborative training solutions are also proposed, such as federated learning [77],
which enables distributed model training across IoT devices. This strategy involves sharing
model updates while preserving data privacy [102], effectively reducing the risk of man-in-
the-middle attacks, malware, eavesdropping, and energy theft [141]. Complementing this
approach, split learning is proposed, which divides the model training task between devices
and a server, ensuring privacy by sharing only intermediate representations instead of raw
data or complete models [140]. This method also enhances the efficiency and scalability of
the training process.

4.5.2. Blockchain
Blockchain technology is recommended for integration into IoT systems due to its
numerous advantages. Blockchain replaces traditional data management systems with a
decentralised architecture, enabling direct data transactions without intermediaries [92,138].
This technology can handle large volumes of transactions while simplifying processes
within the system [110,138]. Through smart contracts, transactions can be automated based
on well-defined rules, reducing the need for manual interventions and lowering transaction
costs [88,92,102].
The management of large data volumes can be improved using off-chain data stor-
age, with only data hashes stored on the blockchain [84]. This approach ensures data
integrity without overloading the blockchain. Ethereum-based frameworks can function
as a trapdoor to ensure data confidentiality in IoT systems [85,130]. During the off-chain
data repositioning process, encryption and decentralised operations are employed to main-
tain data privacy. Another blockchain and Trusted Execution Environment (TEE)-based
framework for distributed data sharing and authentication is proposed in the paper at [144].
Blockchain is utilized for on-chain security and access control, while TEE is employed
for off-chain data protection. Furthermore, an SGX-based distributed storage system is
integrated to enhance data integrity, availability, and resilience against rollback attacks.
Thanks to ledgers that record every transaction, traceability is enhanced, fostering
greater trust in the system [86,92,93,102,138]. Frameworks like Hyperledger Fabric reduce
the risk of unauthorised data access by restricting it to authorised nodes only [84].
Blockchain ensures confidentiality, integrity, and availability [87]. Data confidential-
ity is achieved through digital identity encryption methods. Immutable records on the
blockchain prevent unauthorised data modifications, maintaining integrity [88,91,138]. The
decentralised nature of blockchain technology enhances availability, with data stored across
multiple nodes [87].
Blockchain technology has also been proposed to enhance authentication processes
by integrating Zero-Knowledge Proofs. This method ensures privacy without disclosing
sensitive data while verifying the authenticity of OTPs and confirming user identity [96].

4.5.3. Artificial Intelligence

The use of Artificial Intelligence, particularly Artificial Immune Systems, aids in de-
tecting and mitigating malware attacks at the IoT device or gateway level, addressing risks
without requiring extensive resources [141]. AI methods incorporating Differential Privacy
can protect sensitive biometric data by adding controlled noise to the data, mitigating the
risk of data leakage during transfer [141].
Computers 2025, 14, 61 30 of 45

Computers 2025, 14, x FOR PEER REVIEW 31 of 47

AI in IoT facilitates the development of intelligent, adaptive security solutions tai-
lored to the diverse applications of IoT systems [141]. It can also complement blockchain
technology through
throughits itsanalytical
analytical capabilities,
capabilities,particularly
particularlyin handling
in handlinglargelarge
data volumes.
data vol-
AI identifies
umes. patternspatterns
AI identifies and anomalies, assisting in
and anomalies, transaction
assisting validation validation
in transaction within blockchain
within
systems [85]
blockchain Additionally,
systems it can optimise
[85] Additionally, it cantransactions by dynamically
optimise transactions adjustingadjusting
by dynamically parame-
parameters
ters [85]. [85].
AI enhances ML capabilities,
capabilities, and Generative AI can be employed employed to create
create diverse
diverse
datasets for
for training
trainingmodels
modelswhen whenreal-world
real-world datasets
datasets areare limited
limited [89].
[89]. Additionally,
Additionally, it
it can
can simulate
simulate scenarios
scenarios to improve
to improve decision-making
decision-making and predictive
and predictive capabilities,
capabilities, strength-
strengthening
ening both defensive
both defensive and proactive
and proactive security
security strategies
strategies [89,90].
[89,90]. Moreover,
Moreover, this this technology
technology en-
enhances human-device
hances human-device interaction,asasAIAImodels
interaction, modelscan caninterpret
interprethuman
human voice
voice with greater
accuracy [90]. AI can further reduce failure risks by analysing historical data to predict predict
maintenance
maintenance requirements
requirementsfor fordevices
devices[85].
[85].While
Whilethese
theseadvancements
advancementsstrengthen
strengthen IoT
IoTsecu-
se-
rity and
curity functionality,
and functionality, Generative
Generative AIAIalso
alsointroduces
introducesnew newthreats.
threats.Recent
Recent research, [143],
highlights that adversaries can exploit AI to bypass existing security mechanisms, partic-
ularly NN-based IDSs. A novel novel offensive
offensive strategy
strategy called
called Attack
Attack Obfuscation,
Obfuscation, leveraging
leveraging
Conditional
Conditional GANs,
GANs, hashas been
been proposed
proposed to to evade
evade IDS by injecting synthetic traffic traffic designed
to deceive detection
detection algorithms.
algorithms.
AI is also
also pivotal
pivotal in in addressing
addressing challenges
challenges associated
associated with the introduction
introduction of 6G 6G
networks.
networks. It can optimise network performance, deliver personalised personalised services
services in in 6G based
on user
user behaviour,
behaviour,and andenable
enableinnovative
innovative applications
applications such
suchas as
holographic
holographic communication
communica-
and
tion augmented
and augmentedreality [123].[123].
reality

4.5.4.
4.5.4. Edge
Edge Computing
Computing and and Fog
Fog Computing
Computing
Edge computing and fog computing
Edge computing and fog computing cancanenhance
enhancethe the
efficiency of cooperation
efficiency between
of cooperation be-
IoT systems and the Cloud while also improving their security and scalability
tween IoT systems and the Cloud while also improving their security and scalability [63]. [63]. These
technologies involveinvolve
These technologies performing computational
performing processes
computational closer tocloser
processes the data source
to the data[119], al-
source
though this approach
[119], although may resultmay
this approach in higher
resultenergy consumption
in higher for the selected
energy consumption devices
for the [71].
selected
Nevertheless, the proposed solution reduces latency, enabling faster attack
devices [71]. Nevertheless, the proposed solution reduces latency, enabling faster attack detection [64]
and preventing
detection their
[64] and propagation
preventing theirwithin the system
propagation [69,121].
within the system [69,121].
Additionally,
Additionally, these approaches decrease the amount
these approaches decrease the amount of of data
data transmitted
transmitted to to the
the
Cloud, reducing bandwidth requirements [70] and optimising data transmission
Cloud, reducing bandwidth requirements [70] and optimising data transmission across across the
network
the network[123].
[123].
The
The accompanying heatmap,
accompanying heatmap,Figure
Figure9,9,illustrates
illustratesthe
theutilization
utilizationof ofemerging
emergingtechnolo-
technol-
gies within the identified categories.
ogies within the identified categories.

Figure 9.
Figure 9. Heatmap
Heatmap depicting
depicting the
the prevalence
prevalence ofof Emergent
Emergent Technologies
Technologies across
across identified
identified categories
categories
(red—0% prevalence,
(red—0% prevalence, yellow—below
yellow—below20% 20%prevalence,
prevalence,green
greenand
andits
itsvarying
varyingshades
shadesrepresent
representpreva-
prev-
alenceabove
lence above20%,
20%,with
withdarker
darkershades
shadessignifying
signifyinghigher
higherprevalence).
prevalence).

In the context of Risk Management, the reviewed articles do not employ emergent
technologies. Instead, they focus on developing frameworks to achieve standardizations
and risk management models that guide organizations in formulating cybersecurity
Computers 2025, 14, 61 31 of 45

In the context of Risk Management, the reviewed articles do not employ emergent
technologies. Instead, they focus on developing frameworks to achieve standardizations
and risk management models that guide organizations in formulating cybersecurity im-
plementation policies for Internet of Things systems. However, even within this area, the
cognitive capabilities of Machine Learning and Artificial Intelligence could be utilized to
adapt rules based on the application domain, identify latent risks, and update regulations
in response to emerging threats and technological advancements.
The heatmap illustrates that the four emergent technologies exhibit varying levels of
adoption across the identified categories. Machine Learning is most frequently suggested
as a solution for attack detection due to its capabilities in traffic analysis, anomaly detection,
and resource optimisation. Blockchain technology demonstrates its prominence in the data
management and protection category, attributed to its decentralization features and ability
to ensure data integrity and confidentiality. This category also sees significant utilization of
Artificial Intelligence, particularly Generative AI, which can generate necessary conditions,
such as test data for training ML models, validate transactions within blockchain systems,
and support data integrity assurance. Edge and Fog Computing emerge as deployment
suggestions for attack detection systems and solutions for securing networks and facilitating
communication between edge devices and servers. These technologies contribute by
reducing latency, enhancing security, and ensuring efficient network operations.

4.6. Risk Management

The complexity and dynamic nature of the IoT ecosystem have necessitated the devel-
opment of strategies tailored to this context. To address this need, various risk management
methodologies have been proposed. For instance, study [145] outlines several types of risks
and identifies the main frameworks employed in managing them. Similarly, study [146]
conducts a literature review on risk management, also examining the frameworks discussed
in [145]. Both studies highlight vulnerabilities in existing methodologies.
A novel framework for risk management, IOTA-SRM, is introduced in [147] to address
the limitations of current frameworks. This systematic approach manages risk across
different architectural levels within IoT systems. Additionally, the IoTSRM2 proposed
in [148] underscores the necessity for comprehensive solutions. These approaches empha-
size multi-layered cybersecurity strategies, incorporating encryption, machine learning
for threat detection, and blockchain to ensure secure communications. A critical aspect of
these frameworks is the focus on security at various architectural levels within IoT systems,
particularly at the device and network levels, which are highly vulnerable to attacks such
as DDoS and data interception.
Study [150] presents a lightweight dynamic risk assessment approach that integrates
scenario-based simulations. By utilizing synthetic data and threat models, this method
provides a comprehensive understanding of emerging threats in healthcare settings. The
adaptability of such models facilitates continuous risk assessment and mitigation, thereby
enhancing the security and resilience of MIoT infrastructures against evolving cyber threats.
Despite these advancements, existing security policy models often lack the versatility
to integrate comprehensive risk assessments, regulatory compliance, and AI/ML-driven
adaptability. To address this gap, study [151] introduces an adaptive edge security frame-
work that dynamically generates and adjusts security policies for IoT edge devices. This
framework incorporates AI-driven adaptability, conflict resolution mechanisms, and com-
pliance analysis to ensure that security policies remain responsive to emerging threats,
regulatory changes, and variations in device status.
Additionally, a novel vulnerability-oriented risk identification framework addresses
these limitations by employing a structured four-step process that enhances IoT security risk
Computers 2025, 14, 61 32 of 45

assessments [152]. By applying this framework to a smart healthcare system, researchers

successfully identified critical attack scenarios arising from improper security measures,
mobility concerns, and intercommunication vulnerabilities.
Beyond these technical frameworks, regulatory measures also play a crucial role in
IoT cybersecurity. The Cyber Resilience Act, examined in [149], represents a broad yet
necessary regulatory approach to mitigating IoT security risks, though ambiguities in its
provisions could hinder its effectiveness. Key challenges include legal uncertainties in risk
assessment, vague security requirements, and limitations on manufacturers’ responsibility
for vulnerability management. While not a definitive solution, the CRA lays an essential
foundation for enhancing IoT security, with its ultimate success depending on industry
adoption and regulatory clarity.
Table 8 provides an overview of the key challenges and solutions in risk management,
outlining the primary issues and proposed strategies to address them.

Table 8. Key challenges and solutions in risk management.

Challenge Related Challenges Key Threats Solutions

Creating risk assessment models;
Threat modelling;
Identifying threats and Balancing security constraints and devices; Using ML for real-time risk assessment;
managing vulnerabilities, Performing real-time updates; Compliance-oriented frameworks;
Lack of standardisation in
ensuring resilience in Complying with GDPR and IoT-specific IOTA-SRM framework for risk management;
risk management approach
compliance with data regulations while maintaining Lightweight dynamic risk assessment using
protection standards system functionality scenario-based simulations;
Adaptive edge security framework;
Regulatory approaches such as the Cyber Resilience Act;
References [145–152]

5. Discussion
Based on the analysis of the selected articles and the identification of the categories of
topics addressed by them, a bar chart (Figure 10) was constructed to visualize the relative
weight of concern pertaining to the listed categories. It is noteworthy
Computers 2025, 14, x FOR PEER REVIEW 34 that
of 47 articles that
encompassed multiple categories were considered for each category separately.

Figure Articles
10.Articles
Figure 10. classified
classified by identified
by identified categories.
categories.

5.1. Securing Identity Management

This category reflects the imperative to ensure that only authorized entities have ac-
cess to the network. The escalating prevalence of threats targeting identity theft, unau-
thorized system access, and credential theft underscores the urgency of developing robust
Computers 2025, 14, 61 33 of 45

The bar chart analysis demonstrates that three of the six categories—Securing Identity
Management, Attack Detection, and Emergent Technologies—attract the most attention.
Among these, Emergent Technologies has the tallest bar in the chart, reflecting their ex-
tensive adoption in proposed solutions. This widespread utilization is attributed to their
capability to process large datasets, reduce anomaly detection time, and adapt to the
rapidly evolving landscape of threats and attack types specific to IoT systems. Suggested
approaches include employing machine learning techniques for predicting and preventing
attacks, utilizing blockchain technology to enhance security through decentralization, and
deploying Edge and Fog Computing to minimise latency and prevent the propagation of
attacks across systems.
However, integrating emergent technologies presents new challenges, such as vulner-
abilities and resource constraints inherent to IoT devices. Training artificial intelligence
and machine learning models requires substantial computational resources, posing a sig-
nificant obstacle. Furthermore, regulatory issues and ethical dilemmas arise, particularly
when systems must make decisions that may involve trade-offs or sacrifices. Addressing
these challenges will necessitate innovative solutions to maximize the benefits of these
technologies while mitigating their drawbacks.

5.1. Securing Identity Management

This category reflects the imperative to ensure that only authorized entities have access
to the network. The escalating prevalence of threats targeting identity theft, unauthorized
system access, and credential theft underscores the urgency of developing robust identity
protection mechanisms. These mechanisms encompass multi-factor authentication and
stringent access controls. Future implementations may integrate emerging technologies
such as artificial intelligence, blockchain, and machine learning with biometric identification
methods. Such an approach could enhance the accuracy of biometric authentication, fortify
data security, and ensure adaptability to novel threats.

5.2. Attack Detection

Attack detection is a cornerstone of IoT security, crucial for promptly identifying and
mitigating threats to prevent substantial losses for both systems and users. The focus of
this category underscores the significance of real-time system monitoring, adaptability to
evolving threats, incident response capabilities, and resource optimisation. The reviewed
studies emphasize integrating machine learning techniques to develop adaptive detection
systems with enhanced response times and collaborative methods that distribute detection
tasks across system components.

5.3. Communication and Networking and Data Management and Protection

These two categories received equal attention, reflecting their continued importance
in establishing a secure infrastructure.

5.3.1. Communication and Networking

Secure communication and networking are essential to maintaining a reliable flow of
data within IoT systems. The studies reviewed propose protocols tailored for 5G and 6G
networks and methods to integrate AI for faster response times and reduced resource con-
sumption. Dynamic spectrum sharing is suggested as a solution to bandwidth limitations;
however, it introduces challenges such as interference and unauthorised access that need
to be addressed.
Computers 2025, 14, 61 34 of 45

5.3.2. Data Management and Protection

This category focuses on ensuring encryption, implementing backup strategies, and
adhering to data protection regulations to mitigate data breaches that could lead to financial
or reputational damage. The advent of quantum computing poses a significant challenge
to classical encryption systems, as quantum computers could easily break traditional
cryptographic algorithms.

5.4. Risk Management

As a complementary category, risk management underscores the importance of reg-
ulations in building robust cybersecurity methodologies for IoT systems. This domain
has fewer studies, as it is often reactive, requiring the occurrence of specific limitations to
inspire experimentation and the development of effective frameworks. Incorporating ML
and AI into risk management could enable dynamic rule adaptation based on application
domains, identification of latent risks, and updates to guidelines in response to emerging
threats and technological evolution.
Following the analysis of the selected articles based on the identified categories, the
proposed solutions highlight efforts to implement scalable and adaptable attack detection
and prevention systems capable of handling concept drift. Another key focus in the attack
detection field is optimising response time and reducing false alarms. These systems also
need to function effectively in resource-constrained environments with limited compu-
tational capacity. Researchers are experimenting with machine learning techniques to
address these challenges; however, issues such as long model training times and the need
for continuous adaptability remain significant.
In the area of data security and management, several articles propose blockchain-
based frameworks to ensure authorized data access and protect data integrity during
distribution and storage. These frameworks can be enhanced with Edge Computing,
encryption techniques like zero-knowledge proofs, and secure data-sharing protocols.
For the Network and Communication category, solutions address challenges arising
from the diversity of connected devices and spectrum allocation. Proposed approaches
include grouping devices by capacity and coverage, enabling dynamic spectrum sharing,
and developing IoT security standards to secure device firmware.
The Risk Management category also emphasizes the need for standardization. Articles
in this category propose the development of compliance-oriented frameworks, threat
modelling techniques, and risk assessment models.

5.5. Identified Challenges and Limitations of Integrating Emerging Technologies

The analysis of the papers shows that most of the authors proposed emerging
technologies-based solutions. Integrating these technologies introduces new challenges
and limitations that may impede their practical implementation.

5.5.1. Robust ML-Based Frameworks

Elaborating robust security frameworks for real-world applications is the goal of most
papers analysed in this review. There are solutions proposing the use of model training
methods such as Transfer Learning, Incremental Learning, Deep Learning and Federated
Learning to achieve this goal. However, analysing them separately, using them comes with
numerous limitations from the perspective of adaptability, scalability and concept drift.
IoT devices often operate in resource-constrained environments, limiting their ability
to implement robust security measures. For instance, ML models require significant
computational resources for training and operation, which may exceed the capacity of
low-power IoT devices. Smart home appliances could serve as an example of such devices.
Computers 2025, 14, 61 35 of 45

These appliances, ranging from smart thermostats and light bulbs to security cameras
and home assistants, often operate on minimal resources to reduce costs and enhance
energy efficiency.
The extended training times required for many ML models poses a substantial barrier.
Additionally, the dynamic nature of cyber threats—concept drift—necessitates frequent
retraining of models to ensure their effectiveness against evolving attack patterns. This
process is both resource-intensive and time-consuming, often delaying the deployment of
updated models in real-world scenarios. These limitations underscore the need for inno-
vative approaches to optimise training efficiency, such as the learning methods analysed
in Table 9. However, all these methods involve significant computational resources to
achieve high level performance. Moreover, there are issues with continuous training and
transferring or sharing data among heterogenous devices. A framework using all methods
could mitigate the limitations of them, while also engaging new challenges such as:
• Architecture complexity which involves difficult diagnose process, maintenance, opti-
misation and scalability;
• Training pipeline sophistication to keep a stable model behavior;
• Incremental learning could deteriorate pretrained foundation, introducing errors
and vulnerabilities;
• Communication overhead introduced by the need of data exchange between devices
and central server, as well as between source domain and target domain;
• Computational effort persists.

Table 9. Challenges of learning methods implementation in real-world systems.

Learning Method Challenges

Need of closely related source and target domains;
Model performance degradation if knowledge from source is conflicting or not relevant to target domain;
Could inherit vulnerabilities from source domain;
Transfer Learning
Improper adjustment may lead to loss of generalization capabilities;
Adjusting the target involves high computational and memory costs;
Selecting the right source model not to waste computational resources.
Not suitable for systems with large amount of data because of possibility of forgetting issues when new
data is included;
Incremental Learning Could involve accidental model drift degrading model performance;
Unexpected domain changes lead to instability;
Used in resource constrained environments conducts to suboptimal updates.
Vulnerable to adversarial attacks, causing prediction alteration;
Difficulties with distribution shift between training and real-world data;
Deep Learning Need of large amount of labeled training data;
Failures on edge cases;
Training involves substantial computational resources.
Learning based on non-IID devices generated data leads to poor generalization;
Model poisoning caused by an infected device;
Federated Learning Adversarial attacks targeting local or global data;
Data synchronization issues because of different speed of the connected devices;
Computational limitations lead to incorrect model updates.

5.5.2. AI and Blockchain-Based Frameworks

Emergent technologies, including artificial intelligence and blockchain, offer transfor-
mative potential for enhancing IoT security. However, their adoption introduces unique
vulnerabilities that require careful consideration. AI systems, for instance, are susceptible
to adversarial attacks, where malicious inputs are crafted to manipulate the model’s pre-
dictions or decision-making processes. Similarly, blockchain technology, while providing
decentralized and tamper-resistant solutions, remains vulnerable to specific threats such as
51% attacks, in which an entity gains control of the network’s hashing power, potentially
compromising its integrity. Other consensus-based exploits, such as double-spending or
Computers 2025, 14, 61 36 of 45

transaction malleability, further highlight the risks associated with blockchain deployment.
Some of the primary challenges in deploying blockchain technology include the lack of
expertise in this domain, the complexity of architectures such as Hyperledger Fabric, and
the initial configuration, update, and maintenance efforts.
To ensure the secure integration of these technologies into IoT frameworks, it is es-
sential to develop robust defense mechanisms, such as adversarial training for AI and
improved consensus algorithms for blockchain. Additionally, comprehensive risk as-
sessment and continuous monitoring are necessary to anticipate and mitigate potential
vulnerabilities, ensuring the resilience of IoT systems against emerging threats.
Additionally, a comprehensive regulatory framework must be established to define
standards, establish a consensus mechanism, facilitate governmental and policy manage-
ment, and implement data management strategies.
Table 10 provides a comprehensive summary of the discussion section, consolidating
key findings, research trends, and challenges identified after the review process. The
distribution of research focus indicates that Emergent Technologies (thirty-seven articles),
Attack Detection (twenty-six articles), and Identity Management (twenty-one articles) are
the most explored topics, while Risk Management remains underdeveloped (four articles).
This table highlights the critical areas shaping IoT security research and the ongoing
challenges that must be addressed.

Table 10. Discussion section key points summarized.

Focus point Summary Number of Articles

The most studied areas are Emergent Technologies, Attack 51 (Emergent Technologies),
Detection, and Securing Identity Management, 26 (Attack Detection),
Most Addressed Categories
highlighting their significance in IoT security. Risk 28 (Identity Management),
Management is the least explored 8 (Risk Management)
Widely used for attack detection, anomaly detection, and
Emergent Technologies Adoption secure identity management. ML, Blockchain, and AI are 51
the most discussed
Vulnerabilities and resource constraints inherent to IoT
devices, training artificial intelligence and machine
Challenges of Emergent Technologies -
learning models requires substantial computational
resources, and regulatory issues and ethical dilemmas arise
Focuses on preventing unauthorized access and credential
theft through multi-factor authentication, access controls,
Identity Protection and blockchain-based identity management. Future 28
directions suggest biometric authentication with AI for
enhanced security
Highlights the need for real-time monitoring, fast response
times, and adaptability to evolving threats. ML-based
Attack Detection 26
approaches improve accuracy but face issues with concept
drift, false alarms, and resource limitations.
Protocols tailored for 5G and 6G networks, along with AI
integration, are proposed to enhance data flow reliability
Addresses challenges from diverse connected devices and
Secure Communication and Networking 20
spectrum allocation, proposing solutions like grouping
devices by capacity and coverage, dynamic spectrum
sharing, and IoT security standards
Encryption, backup strategies, and compliance with data
protection regulations are emphasized to prevent breaches.
Data Security 12
Post-quantum cryptography is identified as a growing area
of concern.
Building robust cybersecurity methodologies for IoT
systems, incorporating ML and AI for dynamic rule
adaptation and identification of latent risks
Risk Management 8
Emphasizes the need for standardization, proposing
compliance-oriented frameworks, threat modelling
techniques, and risk assessment models
Highlights the need for global standards to govern IoT
security. Future efforts should focus on AI governance,
Regulatory Framework -
blockchain compliance, and adaptive regulations to keep
up with evolving threats.
Computers 2025, 14, 61 37 of 45

6. Conclusions and Future Work

This paper presented a systematic review of the latest IoT security research, aiming to
identify key directions for enhancing both the security and trustworthiness of IoT systems.
The paper starts by identifying and categorising critical aspects of IoT security, specif-
ically focusing on Attack Detection, Communication and Networking, Securing Iden-
tity Management, Data Management and Protection, Risk Management and using Emer-
gent Technologies.
After the conducted analysis from the paper, it can be concluded that attack detec-
tion techniques are increasingly relying on advanced ML and deep learning models for
precise anomaly detection, reduced false positives, and real-time responsiveness. Data
management and protection emphasise dynamic, blockchain-based solutions to secure sen-
sitive information while ensuring scalability. Identity management has advanced through
blockchain and Edge Computing-based multilevel frameworks, cryptographic techniques
such as zero-knowledge proofs, AES, ring signatures, distributed authentication mecha-
nisms, and secure data-sharing protocols. Device identification using time series, Ethereum
Layer 2 roll-ups, mutual authentication, decentralised PKI, one-time pad encryption, sensor-
based verification, ECC-AES encryption, and automated IoT trust transfer further enhances
the security of IoT systems.
Networking and communication challenges, particularly with 5G/6G environments,
are being tackled through dynamic spectrum sharing and secure protocols. Emerging
technologies such as AI and Edge Computing are proving instrumental in adaptive security
measures, offering real-time anomaly detection and resource efficiency. Despite these
advancements, challenges remain, including the need for standardised datasets, robust
evaluation methods, and scalable solutions that can adapt to an expanding attack surface.
To address the identified challenges and further strengthen IoT security, the following
directions are proposed:
1. Quantum-Resistant Cryptography
With the impending rise of quantum computing, the exploration and adoption of
quantum-resistant cryptographic techniques must be prioritized. Algorithms like
lattice-based cryptography, hash-based signatures, and quantum key distribution
could offer robust protection against future threats posed by quantum computers.
2. Data Integrity and Privacy-Preserving Techniques
Secure management of LSTM IoT data should focus on blockchain-based frameworks
for data integrity. Privacy-preserving methods, such as homomorphic encryption
and differential privacy, must be integrated to ensure secure data sharing without
compromising user privacy.
3. System Resilience and Fallback Strategies
Research should focus on developing secure fallback mechanisms to ensure sys-
tem resilience during failures or breaches. Techniques like redundant architectures,
automated recovery protocols, and distributed denial-of-service (DDoS) mitigation
frameworks are essential for reliable IoT deployments.
4. Optimising Resource Management Using AI and ML
Efficient resource allocation in IoT systems remains a pressing challenge, particu-
larly for resource-constrained devices. AI-driven solutions should be explored to
optimise computational efficiency, improve adaptability to evolving threats, and
minimise latency.
5. Policy and Standards Development
The establishment of international standards and regulatory frameworks is crucial
to promote consistency and interoperability across IoT ecosystems. Policymakers,
researchers, and industry stakeholders should collaborate to develop compliance-
Computers 2025, 14, 61 38 of 45

oriented guidelines that address security and privacy concerns. Future policies should
focus on harmonized global compliance, mandatory security baselines, legal account-
ability, and emergent technologies integration. Simultaneously, international standard-
ization bodies should develop adaptive, interoperable security frameworks, advance
post-quantum cryptography adoption, and explore self-healing IoT architectures.
These directions will pave the way for a secure, resilient, and trustworthy IoT ecosys-
tem, ensuring long-term sustainability and public confidence in IoT technologies.
6. Focus on Securing Neglected IoT Devices
Many IoT devices, particularly in smart homes, remain overlooked in terms of security.
Targeted research is needed to develop lightweight security protocols, automated
firmware updates, and user-friendly mechanisms to protect these devices, which often
operate in resource-constrained environments.
7. Interference Mitigation in Dynamic Spectrum Sharing
As dynamic spectrum sharing grows, mitigating interference and unauthorized spec-
trum access is critical. Future research should explore AI-driven spectrum sensing,
cognitive radio techniques, adaptive interference control, and blockchain-based spec-
trum management to enhance secure and efficient spectrum utilization.
Table 11 summarizes the research directions derived from the analysis.

Table 11. Future research directions in IoT security.

Category Future Directions

Develop AI and ML-based techniques to improve real-time anomaly detection and threat prediction
Attack detection
Securing smart home systems with weak credentials
Data management and protection Integrating blockchain and privacy-preserving techniques
Securing identity management Decentralized identity solutions and advanced authentication mechanisms
Quantum-resistant cryptography
Communication and Networking
Interference mitigation strategies in dynamic spectrum sharing
Emergent technologies Optimise resource management using AI and ML
Investigate secure fallback strategies
Risk management
International standards and regulatory framework development

Author Contributions: Conceptualization, H.S. and D.E.P.; methodology, H.S., D.E.P. and R.D.Z.;
resources, H.S. and D.E.P.; writing—original draft preparation, H.S. and D.E.P.; writing—review
and editing, D.E.P., R.D.Z. and H.S.; visualization D.E.P.; supervision, D.E.P. and R.D.Z.; project
administration D.E.P. All authors have read and agreed to the published version of the manuscript.

Funding: This research received no external funding.

Data Availability Statement: Not applicable.

Conflicts of Interest: The authors declare no conflicts of interest.

Abbreviations
The following abbreviations are used in this manuscript:

2FA Two-factor authentication

ACE Associative Cryptographic Encryption
ADWIN Adaptive Windowing
AI Artificial Intelligence
ARF Adaptive Random Forest
CBAM Convolutional Block Attention Module
CoAP The Constrained Application Protocol
CP-ABE Ciphertext-Policy Attribute-Based Encryption
CRA Cyber Resilience Act
Computers 2025, 14, 61 39 of 45

CTGAN Conditional Tabular Generative Adversarial Networks

CVSS Common Vulnerability Scoring System
DAM Distributed Authentication Mechanism
DBO Dung Beetle Optimiser
DDM Deep Drift Model
DDoS Distributed Denial of Service
DNN Deep Neural Network
DT Decision Tree
ECC Elliptic Curve Cryptography
ECC-AES Elliptic Curve Cryptography with Advanced Encryption Standard
EPA Extended Protocol Architecture
FSMFA Firmware-Secure Multi-Factor Authentication
GA Genetic Algorithms
GDPR General Data Protection Regulation
GRU Gated Recurrent Unit
HIDS Host Intrusion Detection Systems
HRA Honest Re-encryption Attacks
ICN-IoT Information-Centric Networking for IoT
ICS Industrial Control Systems
IDS Intrusion Detection Systems
IoT Internet of Things
IOTA-SRM IoT architecture-based Security Risk Management
IoTSRM2 IoT Security Risk Management Strategy Model
IPS Intrusion Prevention Systems
ISO International Organization for Standardization
KNN k-Nearest Neighbours
LPWAN Low-Power Wide-Area Networks
LSTM Long Short-Term Memory
LTE-M Long Term Evolution for Machines
LWE Learning With Errors
MFA Multi-factor authentication
ML Machine Learning
MQTT Message Queuing Telemetry Transport
MUD Manufacturer Usage Description
NB-IoT Narrow Band-Internet of Things
NIDS Network Intrusion Detection Systems
NFT Non-Fungible Token
NIST National Institute of Standards and Technology
OTP One-Time Password
PKI Public Key Infrastructure
PUF Physically Unclonable Function
RF Random Forest
RFC Request For Comments
RFID Radio Frequency Identification
SCADA Supervisory Control and Data Acquisition
SI-AO Self-Improved Aquila Optimiser
SRPs Sampled Randomized Pooling Strategy
SSL-VPN Secure Sockets Layer Virtual Private Network
SVM Support Vector Machine
TCN Temporal Convolutional Network
TEE Trusted Execution Environment
TL Transfer Learning
VGG16 Visual Geometry Group 16 (number of layers with learnable parameters)
VNSFs Virtual Network Security Functions
Computers 2025, 14, 61 40 of 45

References
1. Greengard, S. Internet of Things. In Encyclopedia Britannica; 2024. Available online: https://www.britannica.com/science/
Internet-of-Things (accessed on 3 January 2025).
2. Satyajit, S. State of IoT 2024: Number of Connected IoT Devices Growing 13% to 18.8 Billion Globally; IoT Analytics: Hamburg,
Germany, 2024.
3. Greenberg, A. Hackers Remotely Kill a Jeep on the Highway—With Me in It. Available online: https://www.wired.com/2015/0
7/hackers-remotely-kill-jeep-highway/ (accessed on 17 January 2025).
4. Antonakakis, M.; April, T.; Bailey, M. Understanding the Mirai Botnet. In Proceedings of the 26th USENIX Security Symposium,
Vancouver, BC, Canada, 16 August 2017.
5. Smart, W. Lessons Learned Review of the WannaCry Ransomware Cyber Attack; Department of Health and Social Care: London,
UK, 2018.
6. Brewster, T. Hackers Used a Fish Tank to Breach a Casino’s High-Roller Database. Forbes. 2018. Available online: https:
//www.forbes.com/sites/thomasbrewster/2018/07/19/fish-tank-hack-into-casino/ (accessed on 17 January 2025).
7. Kari, P. Dozens Sue Amazon’s Ring after Camera Hack Leads to Threats and Racial Slurs. The Guardian, 23 December 2020. Avail-
able online: https://www.theguardian.com/technology/2020/dec/23/amazon-ring-camera-hack-lawsuit-threats (accessed on
17 January 2025).
8. Cimpanu, C. Garmin Services and Production Go Down After Ransomware Attack. Available online: https://www.zdnet.com/
article/garmin-services-and-production-go-down-after-ransomware-attack/ (accessed on 17 January 2025).
9. Easterly, J. The Attack on Colonial Pipeline: What We’ve Learned & What We’ve Done Over the Past Two Years. 2023. Available
online: https://www.cisa.gov/news-events/news/attack-colonial-pipeline-what-weve-learned-what-weve-done-over-past-
two-years (accessed on 15 November 2024).
10. Montalbano, E. Breach Exposes Verkada Security Camera Footage at Tesla, Cloudflare. Threatpost 10 March 2021. Available online:
https://threatpost.com/breach-verkada-security-camera-tesla-cloudflare/164635/ (accessed on 17 January 2025).
11. Gartenberg, C. Security Startup Verkada Hack Exposes 150,000 Security Cameras in Tesla Factories, Jails, and More. 2021. Available
online: https://www.theverge.com/2021/3/9/22322122/verkada-hack-150000-security-cameras-tesla-factory-cloudflare-jails-
hospitals (accessed on 15 November 2024).
12. Greenberg, A. A Hacker Tried to Poison a Florida City’s Water Supply, Officials Say. Available online: https://www.wired.com/
story/oldsmar-florida-water-utility-hack/ (accessed on 17 January 2025).
13. Kapko, M. MOVEit Liabilities Mount for Progress Software. Cybersecurity Dive. 2024. Available online: https://www.
cybersecuritydive.com/news/moveit-liabilities-progress/706015/ (accessed on 17 January 2025).
14. Ptrosyan, A. Annual Number of Internet of Things (IoT) Malware Attacks Worldwide from 2018 to 2022. Statista, Cyber Crime &
Security. 2024. Available online: https://www.statista.com/statistics/1377569/worldwide-annual-internet-of-things-attacks/
(accessed on 17 January 2025).
15. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the Protection of Natural Persons
with Regard to the Processing of Personal Data and on the Free Movement of Such Data, and Repealing Directive 95/46/EC
(General Data Protection Regulation). Off. J. Eur. Union 2016, L119, 1–88.
16. H.R.1668—IoT Cybersecurity Improvement Act of 2020. Available online: https://www.congress.gov/bill/116th-congress/
house-bill/1668 (accessed on 17 January 2025).
17. AT&T, IBM, Nokia, Palo Alto Networks, Symantec and Trustonic Form IoT Cybersecurity Alliance. 2017. Available online:
https://about.att.com/story/iot_cybersecurity_alliance.html (accessed on 1 February 2025).
18. Caindec, K.; Buchheit, M.; Zarkout, B.; Schrecker, S.; Hirsch, F.; Dungana, I.; Martin, R.; Tseng, M. An Industry IoT Foundational
Publication; AT&T Inc.: Dallas, TX, USA, 2017.
19. CoAP RFC 7252 Constrained Application Protocol. Available online: https://datatracker.ietf.org/doc/html/rfc7252 (accessed on
15 November 2024).
20. ISO/IEC 30141:2024; Internet of Things (IoT)—Reference Architecture. International Organization for Standardization: Geneva,
Switzerland, 2024.
21. ETSI EN 303 645; Cyber Security for Consumer Internet of Things: Baseline Requirements. European Standard. June 2020.
Available online: https://www.etsi.org/deliver/etsi_en/303600_303699/303645/02.01.01_60/en_303645v020101p.pdf (accessed
on 9 January 2025).
22. Dritsas, E.; Trigka, M. A Survey on Cybersecurity in IoT. Future Internet 2025, 17, 30. [CrossRef]
23. Szymoniak, S.; Piatkowski,
˛ J.; Kurkowski, M. Defense and Security Mechanisms in the Internet of Things: A Review. Appl. Sci.
2025, 15, 499. [CrossRef]
24. Singh, N.; Buyya, R.; Kim, H. Securing Cloud-Based Internet of Things: Challenges and Mitigations. Sensors 2024, 25, 79.
[CrossRef]
Computers 2025, 14, 61 41 of 45

25. Krzysztoń, E.; Rojek, I.; Mikołajewski, D. A Comparative Analysis of Anomaly Detection Methods in IoT Networks: An
Experimental Study. Appl. Sci. 2024, 14, 11545. [CrossRef]
26. Alshamsi, O.; Shaalan, K.; Butt, U. Towards Securing Smart Homes: A Systematic Literature Review of Malware Detection
Techniques and Recommended Prevention Approach. Information 2024, 15, 631. [CrossRef]
27. Fatima, M.; Rehman, O.; Rahman, I.M.H.; Ajmal, A.; Park, S.J. Towards Ensemble Feature Selection for Lightweight Intrusion
Detection in Resource-Constrained IoT Devices. Future Internet 2024, 16, 368. [CrossRef]
28. Kikissagbe, B.R.; Adda, M. Machine Learning-Based Intrusion Detection Methods in IoT Systems: A Comprehensive Review.
Electronics 2024, 13, 3601. [CrossRef]
29. Dritsas, E.; Trigka, M. Machine Learning for Blockchain and IoT Systems in Smart Cities: A Survey. Future Internet 2024, 16, 324.
[CrossRef]
30. Roy, S.; Sankaran, S.; Zeng, M. Green Intrusion Detection Systems: A Comprehensive Review and Directions. Sensors 2024, 24,
5516. [CrossRef]
31. Alkhayyal, M.; Mostafa, A. Recent Developments in AI and ML for IoT: A Systematic Literature Review on LoRaWAN Energy
Efficiency and Performance Optimization. Sensors 2024, 24, 4482. [CrossRef] [PubMed]
32. Isong, B.; Kgote, O.; Abu-Mahfouz, A. Insights into Modern Intrusion Detection Strategies for Internet of Things Ecosystems.
Electronics 2024, 13, 2370. [CrossRef]
33. Gelgi, M.; Guan, Y.; Arunachala, S.; Samba Siva Rao, M.; Dragoni, N. Systematic Literature Review of IoT Botnet DDOS Attacks
and Evaluation of Detection Techniques. Sensors 2024, 24, 3571. [CrossRef]
34. Rafique, S.H.; Abdallah, A.; Musa, N.S.; Murugan, T. Machine Learning and Deep Learning Techniques for Internet of Things
Network Anomaly Detection—Current Research Trends. Sensors 2024, 24, 1968. [CrossRef]
35. Bukhowah, R.; Aljughaiman, A.; Rahman, M.M.H. Detection of DoS Attacks for IoT in Information-Centric Networks Using
Machine Learning: Opportunities, Challenges, and Future Research Directions. Electronics 2024, 13, 1031. [CrossRef]
36. Alhamarneh, R.A.; Mahinderjit Singh, M. Strengthening Internet of Things Security: Surveying Physical Unclonable Functions
for Authentication, Communication Protocols, Challenges, and Applications. Appl. Sci. 2024, 14, 1700. [CrossRef]
37. Hossain, M.; Kayas, G.; Hasan, R.; Skjellum, A.; Noor, S.; Islam, S.M.R. A Holistic Analysis of Internet of Things (IoT) Security:
Principles, Practices, and New Perspectives. Future Internet 2024, 16, 40. [CrossRef]
38. AlSalem, T.; Almaiah, M.; Lutfi, A. Cybersecurity Risk Analysis in the IoT: A Systematic Review. Electronics 2023, 12, 3958.
[CrossRef]
39. Alotaibi, B. A Survey on Industrial Internet of Things Security: Requirements, Attacks, AI-Based Solutions, and Edge Computing
Opportunities. Sensors 2023, 23, 7470. [CrossRef] [PubMed]
40. Alahmadi, A.A.; Aljabri, M.; Alhaidari, F.; Alharthi, D.J.; Rayani, G.E.; Marghalani, L.A.; Alotaibi, O.B.; Bajandouh, S.A. DDoS
Attack Detection in IoT-Based Networks Using Machine Learning Models: A Survey and Research Directions. Electronics 2023,
12, 3103. [CrossRef]
41. Chui, K.T.; Gupta, B.B.; Liu, J.; Arya, V.; Nedjah, N.; Almomani, A.; Chaurasia, P. A Survey of Internet of Things and Cyber-
Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions. Information 2023, 14, 388.
[CrossRef]
42. Pritika, P.; Shanmugam, B.; Azam, S. Risk Assessment of Heterogeneous IoMT Devices: A Review. Technologies 2023, 11, 31.
[CrossRef]
43. Aslan, Ö.; Aktuğ, S.S.; Ozkan-Okay, M.; Yilmaz, A.A.; Akin, E. A Comprehensive Review of Cyber Security Vulnerabilities,
Threats, Attacks, and Solutions. Electronics 2023, 12, 1333. [CrossRef]
44. Taherdoost, H. Security and Internet of Things: Benefits, Challenges, and Future Perspectives. Electronics 2023, 12, 1901. [CrossRef]
45. Tariq, U.; Ahmed, I.; Bashir, A.K.; Shaukat, K. A Critical Cybersecurity Analysis and Future Research Directions for the Internet
of Things: A Comprehensive Review. Sensors 2023, 23, 4117. [CrossRef]
46. Sun, P.; Wan, Y.; Wu, Z.; Fang, Z.; Li, Q. A Survey on Privacy and Security Issues in IoT-Based Environments: Technologies,
Protection Measures and Future Directions. Comput. Secur. 2025, 148, 104097. [CrossRef]
47. Kumar, S.; Kumar, D.; Dangi, R.; Choudhary, G.; Dragoni, N.; You, I. A Review of Lightweight Security and Privacy for
Resource-Constrained IoT Devices. Comput. Mater. Contin. 2024, 78, 31–63. [CrossRef]
48. Chaurasia, N.; Kumar, P. A Comprehensive Study on Issues and Challenges Related to Privacy and Security in IoT. e-Prime—Adv.
Electr. Eng. Electron. Energy 2023, 4, 100158. [CrossRef]
49. Narciandi-Rodriguez, D.; Aveleira-Mata, J.; García-Ordás, M.T.; Alfonso-Cendón, J.; Benavides, C.; Alaiz-Moretón, H. A
Cybersecurity Review in IoT 5G Networks. Internet Things 2025, 30, 101478. [CrossRef]
50. Bala, B.; Behal, S. AI Techniques for IoT-Based DDoS Attack Detection: Taxonomies, Comprehensive Review and Research
Challenges. Comput. Sci. Rev. 2024, 52, 100631. [CrossRef]
51. Kumari, P.; Jain, A.K. A Comprehensive Study of DDoS Attacks over IoT Network and Their Countermeasures. Comput. Secur.
2023, 127, 103096. [CrossRef]
Computers 2025, 14, 61 42 of 45

52. Makhdoom, I.; Abolhasan, M.; Franklin, D.; Lipman, J.; Zimmermann, C.; Piccardi, M.; Shariati, N. Detecting Compromised IoT
Devices: Existing Techniques, Challenges, and a Way Forward. Comput. Secur. 2023, 132, 103384. [CrossRef]
53. Unpacking IoT Architecture: Layers and Components Explained. Available online: https://deviceauthority.com/unpacking-iot-
architecture-layers-and-components-explained/ (accessed on 3 December 2024).
54. Domínguez-Bolaño, T.; Campos, O.; Barral, V.; Escudero, C.J.; García-Naya, J.A. An Overview of IoT Architectures, Technologies,
and Existing Open-Source Projects. Internet Things 2022, 20, 100626. [CrossRef]
55. Rai, S. How to Greatly Improve Battery Power Efficiency for IoT Devices, Analog Devices, Technical Articles, 6 March 2023.
Available online: https://www.analog.com/en/resources/technical-articles/greatly-improve-battery-power-efficiency-for-iot-
devices.html (accessed on 18 January 2025).
56. Borres, B.; Tenorio, N. How Integrated On/Off Controllers Contribute to Energy Efficient System Designs. Available online: https:
//www.analog.com/en/resources/analog-dialogue/articles/integrated-on-off-controllers-contribute-to-energy-eff.html (ac-
cessed on 1 February 2025).
57. Rottleuthner, M.; Schmidt, T.C.; Wählisch, M. Dynamic Clock Reconfiguration for the Constrained IoT and Its Application to
Energy-Efficient Networking. arXiv 2021, arXiv:2102.10353.
58. Tkhir, P. 4 Types of IoT Networks: Overview and Use Cases. 2023. Available online: https://euristiq.com/types-of-iot-networks/
(accessed on 18 January 2025).
59. Alabsi, B.; Anbar, M.; Rihan, S. Conditional Tabular Generative Adversarial Based Intrusion Detection System for Detecting Ddos
and Dos Attacks on the Internet of Things Networks. Sensors 2023, 23, 5644. [CrossRef]
60. Mishra, N.; Pandya, S. Internet of Things Applications, Security Challenges, Attacks, Intrusion Detection, and Future Visions: A
Systematic Review. IEEE Access 2021, 9, 59353–59377. [CrossRef]
61. Garg, U.; Kumar, S.; Mahanti, A. IMTIBOT: An Intelligent Mitigation Technique for IoT Botnets. Future Internet 2024, 16, 212.
[CrossRef]
62. Chen, Z.; Liu, J.; Shen, Y.; Simsek, M.; Kantarci, B.; Mouftah, H.T.; Djukic, P. Machine Learning-Enabled IoT Security: Open Issues
and Challenges Under Advanced Persistent Threats. ACM Comput. Surv. 2023, 55, 105. [CrossRef]
63. Woodiss-Field, A.; Johnstone, M.N.; Haskell-Dowland, P. Examination of Traditional Botnet Detection on IoT-Based Bots. Sensors
2024, 24, 1027. [CrossRef] [PubMed]
64. Beshah, Y.K.; Abebe, S.L.; Melaku, H.M. Drift Adaptive Online DDoS Attack Detection Framework for IoT System. Electronics
2024, 13, 1004. [CrossRef]
65. Altulaihan, E.; Almaiah, M.A.; Aljughaiman, A. Anomaly Detection IDS for Detecting DoS Attacks in IoT Networks Based on
Machine Learning Algorithms. Sensors 2024, 24, 713. [CrossRef] [PubMed]
66. Farraj, A.; Hammad, E. A Physical-Layer Security Cooperative Framework for Mitigating Interference and Eavesdropping
Attacks in Internet of Things Environments. Sensors 2024, 24, 5171. [CrossRef]
67. Li, M.; Dou, Z. Active Eavesdropping Detection: A Novel Physical Layer Security in Wireless IoT. EURASIP J. Adv. Signal Process.
2023, 2023, 119. [CrossRef]
68. Kim, M.; Suh, T. Eavesdropping Vulnerability and Countermeasure in Infrared Communication for IoT Devices. Sensors 2021,
21, 8207. [CrossRef] [PubMed]
69. Moubayed, A. A Complete EDA and DL Pipeline for Softwarized 5G Network Intrusion Detection. Future Internet 2024, 16, 331.
[CrossRef]
70. Kilichev, D.; Turimov, D.; Kim, W. Next–Generation Intrusion Detection for IoT EVCS: Integrating CNN, LSTM, and GRU Models.
Mathematics 2024, 12, 571. [CrossRef]
71. Abdelhamid, S.; Hegazy, I.; Aref, M.; Roushdy, M. Attention-Driven Transfer Learning Model for Improved IoT Intrusion
Detection. BDCC 2024, 8, 116. [CrossRef]
72. Chen, J.; Xiao, J.; Xu, J. VGGIncepNet: Enhancing Network Intrusion Detection and Network Security through Non-Image-to-
Image Conversion and Deep Learning. Electronics 2024, 13, 3639. [CrossRef]
73. Hu, L.; Zhao, B.; Wang, G. A Network Device Identification Method Based on Packet Temporal Features and Machine Learning.
Appl. Sci. 2024, 14, 7954. [CrossRef]
74. Aroon, N.; Liu, V.; Kane, L.; Li, Y.; Tesfamicael, A.D.; McKague, M. An Architecture of Enhanced Profiling Assurance for IoT
Networks. Electronics 2024, 13, 2832. [CrossRef]
75. Habibi, O.; Chemmakha, M.; Lazaar, M. Imbalanced Tabular Data Modelization Using CTGAN and Machine Learning to Improve
IoT Botnet Attacks Detection. Eng. Appl. Artif. Intell. 2023, 118, 105669. [CrossRef]
76. Alani, M.M. BotStop: Packet-Based Efficient and Explainable IoT Botnet Detection Using Machine Learning. Comput. Commun.
2022, 193, 53–62. [CrossRef]
77. de Caldas Filho, F.L.; Soares, S.C.M.; Oroski, E.; de Oliveira Albuquerque, R.; da Mata, R.Z.A.; de Mendonça, F.L.L.; de Sousa
Júnior, R.T. Botnet Detection and Mitigation Model for IoT Networks Using Federated Learning. Sensors 2023, 23, 6305. [CrossRef]
Computers 2025, 14, 61 43 of 45

78. Negera, W.G.; Schwenker, F.; Debelee, T.G.; Melaku, H.M.; Feyisa, D.W. Lightweight Model for Botnet Attack Detection in
Software Defined Network-Orchestrated IoT. Appl. Sci. 2023, 13, 4699. [CrossRef]
79. Thakkar, A.; Lohiya, R. Attack Classification of Imbalanced Intrusion Data for IoT Network Using Ensemble-Learning-Based
Deep Neural Network. IEEE Internet Things J. 2023, 10, 11888–11895. [CrossRef]
80. Yang, C.; Guan, W.; Fang, Z. IoT Botnet Attack Detection Model Based on DBO-Catboost. Appl. Sci. 2023, 13, 7169. [CrossRef]
81. Hossain, M.A.; Islam, M.S. A Novel Hybrid Feature Selection and Ensemble-Based Machine Learning Approach for Botnet
Detection. Sci. Rep. 2023, 13, 21207. [CrossRef] [PubMed]
82. He, M.; Huang, Y.; Wang, X.; Wei, P.; Wang, X. A Lightweight and Efficient IoT Intrusion Detection Method Based on Feature
Grouping. IEEE Internet Things J. 2024, 11, 2935–2949. [CrossRef]
83. Awajan, A. A Novel Deep Learning-Based Intrusion Detection System for IoT Networks. Computers 2023, 12, 34. [CrossRef]
84. Eghmazi, A.; Ataei, M.; Landry, R.J.; Chevrette, G. Enhancing IoT Data Security: Using the Blockchain to Boost Data Integrity and
Privacy. IoT 2024, 5, 20–34. [CrossRef]
85. Khan, B.U.I.; Goh, K.W.; Khan, A.R.; Zuhairi, M.F.; Chaimanee, M. Integrating AI and Blockchain for Enhanced Data Security in
IoT-Driven Smart Cities. Processes 2024, 12, 1825. [CrossRef]
86. Wei, P.; Wang, D.; Zhao, Y.; Tyagi, S.K.S.; Kumar, N. Blockchain Data-Based Cloud Data Integrity Protection Mechanism. Future
Gener. Comput. Syst. 2020, 102, 902–911. [CrossRef]
87. Jena, S.K.; Barik, R.C.; Priyadarshini, R. A Systematic State-of-Art Review on Digital Identity Challenges with Solutions Using
Conjugation of IOT and Blockchain in Healthcare. Internet Things 2024, 25, 101111. [CrossRef]
88. Song, Z.; Yan, E.; Song, J.; Jiang, R.; Yu, Y.; Chen, T. A Blockchain-Based Digital Identity System with Privacy, Controllability, and
Auditability. Arab. J. Sci. Eng. 2024. [CrossRef]
89. Xu, H.; Li, Y.; Balogun, O.; Wu, S.; Wang, Y.; Cai, Z. Security Risks Concerns of Generative AI in the IoT. IEEE Internet Things Mag.
2024, 7, 62–67. [CrossRef]
90. Wang, X.; Wan, Z.; Hekmati, A.; Zong, M.; Alam, S.; Zhang, M.; Krishnamachari, B. IoT in the Era of Generative AI: Vision and
Challenges. arXiv 2024, arXiv:2401.01923.
91. Wang, F.; Gai, Y.; Zhang, H. Blockchain User Digital Identity Big Data and Information Security Process Protection Based on
Network Trust. J. King Saud. Univ.—Comput. Inf. Sci. 2024, 36, 102031. [CrossRef]
92. Yang, Z.; Liu, Y.; Jin, X.; Luo, X.; Xu, Y.; Li, M.; Chen, P.; Tang, B.; Lin, B. BDIDA-IoT: A Blockchain-Based Decentralized Identity
Architecture Enhances the Efficiency of IoT Data Flow. Appl. Sci. 2024, 14, 1807. [CrossRef]
93. Maeng, J.; Heo, Y.; Joe, I. Hyperledger Fabric-Based Lightweight Group Management (H-LGM) for IoT Devices. IEEE Access 2022,
10, 56401–56409. [CrossRef]
94. Mohammed, M.A.; Wahab, H.B.A. Enhancing IoT Data Security with Lightweight Blockchain and Okamoto Uchiyama Homo-
morphic Encryption. Comput. Model. Eng. Sci. 2024, 138, 1731–1748. [CrossRef]
95. Fan, S.; Wang, J. Multi-Dimension-Precision Chaotic Encryption Mechanism for Internet of Things. Internet Things 2024, 26, 101202.
[CrossRef]
96. Jose Diaz Rivera, J.; Muhammad, A.; Song, W.-C. Securing Digital Identity in the Zero Trust Architecture: A Blockchain Approach
to Privacy-Focused Multi-Factor Authentication. IEEE Open J. Commun. Soc. 2024, 5, 2792–2814. [CrossRef]
97. Bojič Burgos, J.; Pustišek, M. Decentralized IoT Data Authentication with Signature Aggregation. Sensors 2024, 24, 1037. [CrossRef]
[PubMed]
98. Saideh, M.; Jamont, J.-P.; Vercouter, L. Opportunistic Sensor-Based Authentication Factors in and for the Internet of Things.
Sensors 2024, 24, 4621. [CrossRef] [PubMed]
99. Munshi, A.; Alshawi, B. Hybrid Encryption Model for Secured Three-Phase Authentication Protocol in IoT. J. Sens. Actuator Netw.
2024, 13, 41. [CrossRef]
100. Tun, N.W.; Mambo, M. Secure PUF-Based Authentication Systems. Sensors 2024, 24, 5295. [CrossRef]
101. Zhang, B.; Zhang, T.; Xi, Z.; Chen, P.; Wei, J.; Liu, Y. Secure Device-to-Device Communication in IoT: Fuzzy Identity from Wireless
Channel State Information for Identity-Based Encryption. Electronics 2024, 13, 984. [CrossRef]
102. Wang, J.; Li, J. Blockchain and Access Control Encryption-Empowered IoT Knowledge Sharing for Cloud-Edge Orchestrated
Personalized Privacy-Preserving Federated Learning. Appl. Sci. 2024, 14, 1743. [CrossRef]
103. Fenner, J.; Galeas, P.; Escobar, F.; Neira, R. Secure IoT Communication: Implementing a One-Time Pad Protocol with True Random
Numbers and Secure Multiparty Sums. Appl. Sci. 2024, 14, 5354. [CrossRef]
104. Höglund, J.; Bouget, S.; Furuhed, M.; Preuß Mattsson, J.; Selander, G.; Raza, S. AutoPKI: Public Key Infrastructure for IoT with
Automated Trust Transfer. Int. J. Inf. Secur. 2024, 23, 1859–1875. [CrossRef]
105. El-Hajj, M.; Beune, P. Decentralized Zone-Based PKI: A Lightweight Security Framework for IoT Ecosystems. Information 2024,
15, 304. [CrossRef]
106. Zhang, J.; Ouda, A.; Abu-Rukba, R. Authentication and Key Agreement Protocol in Hybrid Edge–Fog–Cloud Computing
Enhanced by 5G Networks. Future Internet 2024, 16, 209. [CrossRef]
Computers 2025, 14, 61 44 of 45

107. Baird, I.; Ghaleb, B.; Wadhaj, I.; Russell, G.; Buchanan, W.J. Securing IoT: Mitigating Sybil Flood Attacks with Bloom Filters and
Hash Chains. Electronics 2024, 13, 3467. [CrossRef]
108. Zerrouki, F.; Ouchani, S.; Bouarfa, H. PUF-Based Mutual Authentication and Session Key Establishment Protocol for IoT Devices.
J. Ambient. Intell. Humaniz. Comput. 2023, 14, 12575–12593. [CrossRef]
109. Nimmy, K.; Sankaran, S.; Achuthan, K. A Novel Lightweight PUF Based Authentication Protocol for IoT without Explicit CRPs in
Verifier Database. J. Ambient. Intell. Humaniz. Comput. 2023, 14, 6227–6242. [CrossRef]
110. Ramachandraiah, K.R.D.; Bommagani, N.J.; Jayapal, P.K. Enhancing Healthcare Data Security in IoT Environments Using
Blockchain and DCGRU with Twofish Encryption. Inf. Dyn. Appl. 2023, 2, 173–185. [CrossRef]
111. Zhang, S.; Du, X.; Liu, X. A Novel and Quantum-Resistant Handover Authentication Protocol in IoT Environment. Wirel. Netw.
2023, 29, 2873–2890. [CrossRef]
112. Chen, Z.; Cheng, Z.; Luo, W.; Ao, J.; Liu, Y.; Sheng, K.; Chen, L. FSMFA: Efficient Firmware-Secure Multi-Factor Authentication
Protocol for IoT Devices. Internet Things 2023, 21, 100685. [CrossRef]
113. Román, R.; Arjona, R.; Baturone, I. A Quantum-Safe Authentication Scheme for IoT Devices Using Homomorphic Encryption
and Weak Physical Unclonable Functions with No Helper Data. Internet Things 2024, 28, 101389. [CrossRef]
114. Rehman, M.U.; Shafqiue, A. Robust Encryption Framework for IoT Devices Based on Bit-Plane Extraction, Chaotic Sine Models,
and Quantum Operations. Internet Things 2024, 27, 101241. [CrossRef]
115. Hou, J.; Peng, C.; Tan, W. A Lattice-Based Data Sharing Functional Encryption Scheme with HRA Security for IoT. Expert. Syst.
Appl. 2024, 254, 124355. [CrossRef]
116. Deng, W.; Li, J.; Yan, H.; Voundi Koe, A.S.; Huang, T.; Wang, J.; Peng, C. Self-Sovereign Identity Management in Ciphertext Policy
Attribute Based Encryption for IoT Protocols. J. Inf. Secur. Appl. 2024, 86, 103885. [CrossRef]
117. Gasmi, M.; Kerdoudi, M.L.; Bachir, A. Load-Balanced Attribute-Based Outsourced Encryption for Constrained IoT Devices.
Comput. Electr. Eng. 2024, 118, 109424. [CrossRef]
118. Velmurugan, P.; Senthil kumar, K.; Sridhar, S.S.; Gotham, E. An Advanced and Effective Encryption Methodology Used for
Modern IoT Security. Mater. Today Proc. 2023, 81, 389–394. [CrossRef]
119. Achkouty, F.; Gallon, L.; Chbeir, R. RDSC: Range-Based Device Spatial Clustering for IoT Networks. Sensors 2024, 24, 5851.
[CrossRef]
120. Ehmer, J.; Savaria, Y.; Granado, B.; David, J.-P.; Denoulet, J. Network Attack Classification with a Shallow Neural Network for
Internet and Internet of Things (IoT) Traffic. Electronics 2024, 13, 3318. [CrossRef]
121. Canavese, D.; Mannella, L.; Regano, L.; Basile, C. Security at the Edge for Resource-Limited IoT Devices. Sensors 2024, 24, 590.
[CrossRef]
122. Singh, C.; Kumar, M.; Upadhyay, M.; Chauhan, P.; Sharma, M. A 6G Network: Future of Nations? Challenges in 6G Communica-
tions. Tuijin Jishu/J. Propuls. Technol. 2023, 44, 73–76.
123. Maduranga, M.W.P.; Tilwari, V.; Rathnayake, R.M.M.R.; Sandamini, C. AI-Enabled 6G Internet of Things: Opportunities, Key
Technologies, Challenges, and Future Directions. Telecom 2024, 5, 804–822. [CrossRef]
124. Bakhshi, T.; Ghita, B.; Kuzminykh, I. A Review of IoT Firmware Vulnerabilities and Auditing Techniques. Sensors 2024, 24, 708.
[CrossRef]
125. Al Hanif, A.; Ilyas, M. Effective Feature Engineering Framework for Securing MQTT Protocol in IoT Environments. Sensors 2024,
24, 1782. [CrossRef]
126. Long, Z.; Yan, H.; Shen, G.; Zhang, X.; He, H.; Cheng, L. A Transformer-Based Network Intrusion Detection Approach for Cloud
Security. J. Cloud Comput. 2024, 13, 5. [CrossRef]
127. Rajan, D.M.; Aravindhar, D.J. Detection and Mitigation of DDOS Attack in SDN Environment Using Hybrid CNN-LSTM. Migr.
Lett. 2023, 20, 407–419. [CrossRef]
128. Sarhan, M.; Layeghy, S.; Moustafa, N.; Gallagher, M.; Portmann, M. Feature Extraction for Machine Learning-Based Intrusion
Detection in IoT Networks. Digit. Commun. Netw. 2024, 10, 205–216. [CrossRef]
129. Hu, Z. Knowledge Graph Based Large Scale Network Security Threat Detection Techniques. Appl. Math. Nonlinear Sci. 2024, 9.
[CrossRef]
130. Oktian, Y.E.; Le, T.-T.-H.; Jo, U.; Laksmono, A.M.A.; Kim, H. Secure Decentralized Firmware Update Delivery Service for Internet
of Things. Internet Things 2024, 26, 101136. [CrossRef]
131. Nguyen, H.D.; Le Sommer, N.; Mahéo, Y. Over-the-Air Firmware Update in LoRaWAN Networks: A New Module-Based
Approach. Procedia Comput. Sci. 2024, 241, 154–161. [CrossRef]
132. Cheng, Y.; Yang, S.; Lang, Z.; Shi, Z.; Sun, L. VERI: A Large-Scale Open-Source Components Vulnerability Detection in IoT
Firmware. Comput. Secur. 2023, 126, 103068. [CrossRef]
133. Verderame, L.; Ruggia, A.; Merlo, A. PARIOT: Anti-Repackaging for IoT Firmware Integrity. J. Netw. Comput. Appl. 2023,
217, 103699. [CrossRef]
Computers 2025, 14, 61 45 of 45

134. Kaushik, K.; Bhardwaj, A.; Dahiya, S. Framework to Analyze and Exploit the Smart Home IoT Firmware. Meas. Sens. 2025,
37, 101406. [CrossRef]
135. Xu, J.; Zhaojun, X.; Wenli, Y.; Hu, W.; Cabani, A.; Xinrong, H. An Intelligent Mechanism for Dynamic Spectrum Sharing in 5G IoT
Networks. Expert Syst. Appl. 2024, 252, 124122. [CrossRef]
136. Alkhaldi, T.M.; Darem, A.A.; Alhashmi, A.A.; Al-Hadhrami, T.; Osman, A.E. Enhancing Smart City IoT Communication: A Two-
Layer NOMA-Based Network with Caching Mechanisms and Optimized Resource Allocation. Comput. Netw. 2024, 255, 110857.
[CrossRef]
137. Ortiz-Ruiz, E.; Bermejo, J.R.; Sicilia, J.A.; Bermejo, J. Machine Learning Techniques for Cyberattack Prevention in IoT Systems: A
Comparative Perspective of Cybersecurity and Cyberdefense in Colombia. Electronics 2024, 13, 824. [CrossRef]
138. Valencia-Arias, A.; González-Ruiz, J.D.; Verde Flores, L.; Vega-Mori, L.; Rodríguez-Correa, P.; Sánchez Santos, G. Machine
Learning and Blockchain: A Bibliometric Study on Security and Privacy. Information 2024, 15, 65. [CrossRef]
139. El-Sofany, H.; El-Seoud, S.A.; Karam, O.H.; Bouallegue, B. Using Machine Learning Algorithms to Enhance IoT System Security.
Sci. Rep. 2024, 14, 12077. [CrossRef]
140. Priyadarshini, I. Anomaly Detection of IoT Cyberattacks in Smart Cities Using Federated Learning and Split Learning. Big Data
Cogn. Comput. 2024, 8, 21. [CrossRef]
141. Alrubayyi, H.; Alshareef, M.S.; Nadeem, Z.; Abdelmoniem, A.M.; Jaber, M. Security Threats and Promising Solutions Arising
from the Intersection of AI and IoT: A Study of IoMT and IoET Applications. Future Internet 2024, 16, 85. [CrossRef]
142. Tekin, N.; Acar, A.; Aris, A.; Uluagac, A.S.; Gungor, V.C. Energy Consumption of On-Device Machine Learning Models for IoT
Intrusion Detection. Internet Things 2023, 21, 100670. [CrossRef]
143. Coppolino, L.; D’Antonio, S.; Mazzeo, G.; Uccello, F. The Good, the Bad, and the Algorithm: The Impact of Generative AI on
Cybersecurity. Neurocomputing 2025, 623, 129406. [CrossRef]
144. Xie, H.; Zheng, J.; He, T.; Wei, S.; Hu, C. TEBDS: A Trusted Execution Environment-and-Blockchain-Supported IoT Data Sharing
System. Future Gener. Comput. Syst. 2023, 140, 321–330. [CrossRef]
145. Kandasamy, K.; Srinivas, S.; Achuthan, K.; Rangan, V.P. IoT Cyber Risk: A Holistic Analysis of Cyber Risk Assessment
Frameworks, Risk Vectors, and Risk Ranking Process. EURASIP J. Inf. Secur. 2020, 2020, 8. [CrossRef]
146. Parsons, E.K.; Panaousis, E.; Loukas, G.; Sakellari, G. A Survey on Cyber Risk Management for the Internet of Things. Appl. Sci.
2023, 13, 9032. [CrossRef]
147. Affia, A.O.; Nolte, A.; Matulevičius, R. IoT Security Risk Management: A Framework and Teaching Approach. Inform. Educ. 2023,
22, 555–588. [CrossRef]
148. Popescu, T.; Popescu, A.; Prostean, G. IoT Security Risk Management Strategy Reference Model (IoTSRM2). Future Internet 2021,
13, 148. [CrossRef]
149. Shaffique, M.R. Cyber Resilience Act 2022: A Silver Bullet for Cybersecurity of IoT Devices or a Shot in the Dark? Comput. Law
Secur. Rev. 2024, 54, 106009. [CrossRef]
150. Czekster, R.M.; Webber, T.; Furstenau, L.B.; Marcon, C. Dynamic Risk Assessment Approach for Analysing Cyber Security Events
in Medical IoT Networks. Internet Things 2025, 29, 101437. [CrossRef]
151. Halgamuge, M.N.; Niyato, D. Adaptive Edge Security Framework for Dynamic IoT Security Policies in Diverse Environments.
Comput. Secur. 2025, 148, 104128. [CrossRef]
152. Beyrouti, M.; Lounis, A.; Lussier, B.; Bouabdallah, A.; Samhat, A.E. Vulnerability-Oriented Risk Identification Framework for IoT
Risk Assessment. Internet Things 2024, 27, 101333. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.