깣

⬛ This three-layer design is modular and efficient. Each layer has

’
devices suited to its role, making it easy to manage and scale.
4. ETHERNET
4.1 Introduction

• Ethernet was developed by Xerox PARC (Palo Alto Research Center) in

the 1970s.

• Bob Metcalfe and D.R. Boggs introduced Ethernet in 1972.

• The first successful data transmission using Ethernet occurred in

1G76 at a speed of 3 Mbps.

• In 1G80, the IEEE defined the Ethernet standard as IEEE 802.3.

• Fast Ethernet (1990s) improved performance without replacing existing

cabling.

4.2 Evolution and Role of Ethernet

• Ethernet is a dominant wired LAN technology used in:

o Homes
o Offices
o Enterprises
o Data centers
o Wide Area Networks (WANs)

• Ethernet works well with Wi-Fi and mobile networks (4G/5G) for
multimedia needs.

4.3 Applications of Ethernet

a) Home Networking

• Used for local home networks to connect PCs and internet.

• Wi-Fi has reduced reliance on Ethernet but Ethernet is still used.

• Powerline Carrier (PLC) and Power over Ethernet (PoE) have extended Ethernet
 use at home:

o PLC: Uses existing electric wiring to transmit Ethernet signals.

o PoE: Provides power through Ethernet cables (e.g., for TVs, PCs).
b) Office Networking

• Replaced older technologies like Token Ring and FDDI.

• Ethernet is:
o Fast
o Simple
o Secure
• Combines with Wi-Fi to support both mobile and fixed devices.
c) Enterprise Networking

• Scalable for:
o High-speed server connections
o LAN backbones
o Campus-wide links
• Supports mixed speeds (10 Mbps to 100 Gbps) using the same protocol.
d) Data Centers

• Dominates in modern data centers.

• Replaced older high-speed tech like InfiniBand and Fiber Channel.

• Technologies include:
1. Key Features of Cloud Computing (According to NIST):

• Broad Network Access: You can use cloud services on many devices
like phones, laptops, tablets.

• Rapid Elasticity: You can quickly get more resources when needed and
reduce them when not needed.

• Measured Service: Cloud tracks how much resources you use so you pay
 only for what you use.

• On-Demand Self-Service: You can start or stop services anytime

without needing help from the provider.

• Resource Pooling: Cloud providers share resources among many

users efficiently, without you worrying where exactly the
resources are.
2. How Cloud Computing Works (Example):

• Companies have their own computers connected to the internet.

• These connect to large servers managed by cloud providers.

• Cloud providers handle things like security and maintenance for these
servers.
3. Benefits of Cloud Computing:

• Flexibility:

o Scale up or down your resources easily.

o Choose different storage types (public, private, hybrid).
o Select services like software, platforms, or infrastructure.
o Use security tools to keep data safe.
• Efficiency:

o Access your data and apps from anywhere with internet.

o Launch applications faster without worrying about hardware.
o Data is backed up, so hardware failure doesn’t mean data loss.
o Pay only for what you use, saving money.
• Strategic Value:

o Providers manage the technology so companies can focus on their

work.
o Regular updates give users the latest technology.
o Teams can collaborate worldwide.
 o Companies stay competitive by using the best tech easily.
4. Cloud Networking:

• This means the networks and connections needed to make cloud

computing work.

• Sometimes cloud data doesn’t travel over the public internet but uses
private secure networks.

• Network security like firewalls protect access to cloud services.

5. Cloud Storage:

• Cloud storage is a part of cloud computing focused on storing data.

• It allows users and small businesses to save data remotely without

buying and managing their own storage.

• Storage can grow or shrink depending on user needs.

1.8 Internet of Things (IoT)

The Internet of Things (IoT) refers to the growing network of smart
devices connected to the Internet, ranging from everyday appliances to
tiny sensors. These devices communicate with people and each other,
often through cloud systems, to collect data, interact with their
environment, and sometimes modify their behavior to manage larger
systems like factories or cities.
1.8.1 Things on the IoT

• IoT devices embed short-range mobile transceivers into gadgets and items.

• They send sensor information, respond to environments, and

sometimes self- modify.

• Devices vary in bandwidth needs: some (like video cameras and VoIP
phones) require high bandwidth, while many others only send small
data packets intermittently.
Ǫuality of Service (ǪoS)

• Definition:
ǪoS refers to the technical parameters of a telecommunications
service that affect how well it meets the needs of users. It focuses on
measurable network and system characteristics.

• Background:
Originally, the Internet was designed as a "best-effort" network without
guarantees on performance. ǪoS architectures like Integrated Services
(IntServ) and Differentiated Services (DiffServ) were developed to
support high-quality real-time services such as VoIP and video
streaming.

• Technology-Centered Approach:
ǪoS is mainly defined from a technical perspective (by ITU standards).
ITU Recommendation E.800 defines ǪoS as the totality of characteristics
that impact the user’s ability to meet their communication needs.
However, in practice, ǪoS focuses on technical metrics rather than direct user
satisfaction.

• Common ǪoS Metrics:

o Throughput: Data transfer rate (bytes/sec or bits/sec)

o Delay (Latency): Time taken for data to travel through the network
o Packet Jitter: Variation in delay between packets
o Error Rate: Fraction of bits with errors
o Packet Loss: Fraction of lost packets
o Priority: Levels assigned to traffic flows affecting handling order
o Availability: Percentage of time the service is operational
o Security: Levels or types of security offered
Ǫuality of Experience (ǪoE)

• Definition:
ǪoE measures the user’s overall performance and satisfaction
with an ICT service or product, combining both objective and
subjective psychological measures.
• Focus:
ǪoE shifts focus from technical parameters to the user’s perception and
actual experience while using the service.

• Components:

o Objective psychological measures: Task completion time,

accuracy, efficiency (do not rely on user opinion)

o Subjective psychological measures: User’s perceived quality

and satisfaction (based on opinions)

• Context Sensitivity:
ǪoE depends on both the technical quality (ǪoS) and the usage
context (e.g., type of communication task). The same ǪoS level
can yield different ǪoE outcomes in different situations.
• Challenges:

o Converting ǪoE features into quantitative metrics

o Mapping ǪoE metrics back to ǪoS parameters to manage and
improve user experience effectively
 Congestion Control

◆ What is Congestion?

• Happens when too many packets are sent, more than network capacity.
• Causes poor performance and reduced quality of service (ǪoS).

2.6.1 Effects of Congestion

1. Ǫueueing Delay:

o Time packets wait in a router’s buffer before being sent.

o Happens when packets arrive faster than they can be processed.
2. Packet Loss:

o Some packets are dropped and don’t reach the destination.

o Caused by errors or too much traffic.
o Affects real-time apps like video calls, games.
3. Blocking of New Connections:

o New connections may be denied due to lack of resources.

2.6.2 Congestion Control Techniques

A. Open Loop Control (Prevent congestion before it happens)

1. Retransmission Policy:

o Avoid unnecessary retransmissions.

2. Window Policy:

o Use Selective Repeat instead of Go-back-N to avoid duplicate

packets.
3. Discarding Policy:

o Drop less important packets (e.g. in audio).

4. Acknowledgment Policy:

o Send fewer acknowledgments (e.g. one ACK for every N packets).

5. Admission Policy:

o Block new connections if congestion is likely.

B. Closed Loop Control (Fix congestion after it happens)

1. Backpressure:

o Congested node stops accepting packets → slows down upstream

nodes.
2. Choke Packet:

o Router sends special packet to source telling it to slow down.

3. Implicit Signaling:

o No feedback; source guesses congestion due to missing ACKs.

4. Explicit Signaling:

o Routers mark packets to warn source/destination.

Types:

o Forward signaling: Warn destination.

o Backward signaling: Warn source.
Approaches:

o Binary: 1-bit warning.

o Credit-based: Source sends packets based on credits.
o Rate-based: Source follows data rate limit set by network.
Software-Defined Networking (SDN)
⬛

v
z
‘
’ What is SDN?

• SDN is a modern way to manage networks.

• It separates the software that controls the network (control plane) from
the hardware that forwards the data (data plane).

• This gives flexibility, central control, and easier management.

구 Key Features of SDN Architecture:

굯
×˙´군
.
k

1. Directly Programmable – Network can be controlled by software directly.

2. Agile – Can adjust traffic flow easily as per demand.

3. Centrally Managed – A single controller manages the whole network.

4. Programmatically Configured – Admins can write programs to control

the network without depending on any one company.

5. Open and Vendor-Neutral – Works on open standards, not locked to

one company.

☼
O
◦ How SDN Works:

• Earlier: Every router had both control + data functions built-in.

• Now (with SDN):

o A central controller makes all the smart decisions (routing,

security, policies).

o Switches only forward the data based on instructions from the

controller.
o Communication between controller and switches uses
standard protocols (like OpenFlow).

v
• Network Functions Virtualization (NFV)

NFV is about moving network functions (like firewall, router, etc.) from
hardware to software.

• These functions now run as software on virtual machines (VMs) on standard

hardware (like regular servers).

* Examples of Network Functions Moved to Software:

˛C

• Routing

• Firewall
• Intrusion Detection

• NAT (Network Address Translation)

´虊
Key Goals of NFV:

• Use software instead of hardware.

• Run on standard computers, not special network devices.

• Use open APIs for communication.

• Deploy and update network functions easily and quickly.

⬛
Benefits of NFV:

• Saves space, power, and cost.

• Easier to maintain and upgrade.

• Longer hardware life.

SDN Data Plane and OpenFlow

What is SDN?

Software-Defined Networking (SDN) is a modern network architecture that

separates the control plane (brains of the network) from the data plane (actual
packet forwarding). It makes networks more programmable and flexible.

3.4.1 SDN Data Plane

• The data plane is also called the infrastructure layer or resource layer (as per
ITU-T Y.3300).

• It includes the network devices (switches/routers) that forward data packets.

• These devices do not make decisions themselves — they simply follow

the instructions from the SDN controller.

• The SDN controller uses protocols to manage and control these devices.

3.4.2 Functions of the Data Plane

Network devices (switches) in the SDN data plane perform two key
functions:
1. Control Support Function

• Enables communication with the SDN controller.

• Uses OpenFlow protocol to receive instructions (like install/update rules).

2. Data Forwarding Function

• Receives data packets from other devices or networks.

• Forwards packets based on pre-installed rules in forwarding tables.

• Can modify, forward, or drop packets.

• Packets may wait in input/output queues before processing or sending.

• Devices usually have 3 I/O ports:

o 1 for communication with controller.

o 2 or more for data in/out.
3.4.3 Data Plane Protocols

• Data packets typically include IP traffic, but forwarding can also

depend on higher layer headers like TCP or UDP.

• Devices use these headers to decide where to send the packet.

• The southbound API (e.g., OpenFlow) is used for communication

between controller and device.

• OpenFlow Protocol Data Units (PDUs) carry instructions from the controller
to the devices.

Key Point: OpenFlow

• OpenFlow is the most popular SDN protocol for communication between

SDN controller and data plane devices.

• It defines:

o How forwarding rules are installed.

o How packets should be processed.

OpenFlow Logical Network Devices and SDN (Software Defined Networking

⬛ What is OpenFlow in SDN?

• OpenFlow is the first SDN protocol, allowing a controller to manage the

forwarding plane (packet routing) of switches/routers.

• It is defined by the Open Networking Foundation (ONF).

• It connects SDN controllers with network devices securely via TLS
(Transport Layer Security).

껩 Components of OpenFlow Network

1. SDN Controller – The brain that tells devices what to do.

2. OpenFlow Switch – Follows instructions from the controller.

3. OpenFlow Channel – A secure interface between the switch and controller.

4. End Systems – Devices like PCs/phones sending or receiving data.

”꺊
Network Services Abstraction Layer (NSAL)

˙•Q What is Abstraction?

• Abstraction hides complex details and shows only necessary information.

• In SDN, it allows applications to focus on what to do, not how to do it.

• Example: Using APIs instead of writing low-level device code.

깣 4.7.1 Abstractions in SDN (As per Scott Shenker – ONF)

’
⬛

SDN is built on 3 main abstractions (see Fig 4.19):

1
⬛ Forwarding Abstraction

• Hides hardware (switch) details from the controller.

• Allows apps to define how packets are forwarded.

• Example: OpenFlow API.

³⬛ Distribution Abstraction

• Deals with multiple distributed controllers.

• Ensures global network view (even in a decentralized setup).

• Helps controllers share and sync network information.

• Example: OpenDaylight or Ryu NOS.

³⬛ Specification Abstraction

• Provides apps with an abstract view to define network goals (e.g., routing or security),
not the steps.

• Hides physical network data.

• Example: A single virtual switch that represents the physical network (see Fig 4.20).

• A MAC learning module can run at the application level to handle unknown hosts
dynamically.
4.13 Information-Centric Networking (ICN)
> Concept

• Focus: Fetch information based on content, not location.

• Difference from traditional networking:

o No need to know server address.

o Network finds the best source of content.

4.13.1 CCNx (Content-Centric Networking)
• Developed by PARC.

• Uses Interest packets (requests) and Content packets (replies).

• Stores previously fetched content in Content Store.

• FIB: For forwarding Interest packets.

• PIT: Tracks requests to route replies correctly.

4.13.2 SDN + ICN with Abstraction Layer

• Challenge: ICN uses content names; SDN uses IP addresses.

• Solution:
o Use a wrapper to map ICN names to SDN-readable fields (like IP).

o Hash content names and use OpenFlow to forward based on hash.

• Modules added to SDN controller:

1. Measurement: Monitors popular content.

2. Optimization: Places popular content optimally in the network.

3. Deflection: Maps popular content to specific ports.

• Benefits:

o Enables ICN features without changing existing SDN switches.

o Efficient caching and resource use via deflection and measurement.

Virtual Machines:

⬛ Traditional Challenges
• Earlier, one OS ran on one PC/server at a time.

• Applications had to be rewritten for different OS/platforms → more time, cost,

and errors.

⬛ Virtualization Concept
• Virtualization = running multiple OSes/applications on one server.

• Achieved by creating Virtual Machines (VMs) – software-based emulation of

physical servers.

⬛ Virtual Machine Monitor (Hypervisor)

• Hypervisor/VMM: Software that allows multiple VMs to share one physical host.

• Acts as a resource manager between hardware and VMs.

• Consolidation ratio: Number of VMs per physical host (e.g., 6:1).

⬛ Benefits of Virtualization
• Fewer physical servers → saves cost, space, power, cables.

• Easier to run legacy apps, increase hardware usage.

• Supports cloud computing and big data efficiently.

⬛ VM as a Software Unit
• A VM includes:

o vCPU, RAM, Storage, NICs, etc.

• Can be created/duplicated quickly by copying files (easy provisioning).

• Faster setup than physical servers (minutes vs months).

⬛ Increased Availability
Definition of Ǫuality of Experience (ǪoE)
Ǫuality:

• It is a user’s judgment or verdict after they perceive, reflect on, describe, and
evaluate an event or service.

• Ǫuality score/rating is usually given on a scale to represent this evaluation.

ǪoE Strategies in Practice

• Studies show that many Ǫuality of Service (ǪoS) parameters affect overall ǪoE.

• This leads to a ǪoE/ǪoS layered approach, where ǪoE and ǪoS are
complementary.

ǪoE/ǪoS Layered Model (Figure 7.13):

1. User level:

o Measures user’s delight or annoyance.

o ǪoE varies individually and can depend on stable or changing personal

factors.

2. Service level:
o Interface between user and service (e.g., video display).

o Measures tolerance thresholds like startup time, buffering,

channel change delays.

3. Application-level ǪoS (AǪoS):

o Controls app-specific parameters: resolution, bitrate, frame rate, codecs,
etc.

o Adjusted based on network bandwidth to maintain quality.

4. Network-level ǪoS (NǪoS):
o Concerns network parameters like coverage, bandwidth, delay, packet
loss.

o Network issues impact ǪoE, especially in interactive services (e.g., VoIP,

7.10 Factors Influencing ǪoE
Ǫuality of Experience (ǪoE) depends on both technical and non-technical factors. Key
factors include:

• User demographics:
Stable traits like attitudes toward new tech, socioeconomic status,
cultural background, and prior knowledge affect perception of quality.

• Type of device:
Different devices (e.g., smart TV vs. smartphone) have different
capabilities, affecting ǪoE differently.

• Content:
Interactive or on-demand content usually results in higher engagement
and better ǪoE compared to linear TV, because users choose what to
watch actively.

• Connection type:
Expectations vary with connection (3G vs. wired). Users tend to be more
tolerant of impairments on mobile/smaller devices.

• Media quality (audio-visual):

Audio and video quality impact ǪoE differently depending on content
type. For simple scenes, audio quality matters more; for high-motion
scenes, video quality is more important.

• Network:
Delays, jitter, packet loss, and bandwidth affect streaming. Interruptions
like rebuffering severely degrade ǪoE and should be minimized even if
startup delay increases.

• Usability:
The ease of using the service impacts ǪoE — less user effort needed
means better ǪoE.
8.2 Classification Models of ǪoE/ǪoS Mapping
ǪoE/ǪoS mapping models are classified by their input type into:

1. Black-Box Media-Based Models

• How they work: Analyze media signals only at system input/output without
internal details.

• Two types:

o Full-reference (double-sided): Compare original (clean) media with

degraded media to measure quality difference perceptually.

o No-reference (one-sided): Use only the degraded media to detect

distortions and estimate ǪoE.

• Pros: Generic, works across systems, privacy-respecting if no original media

needed.

• Cons: Often need access to media signals which may be restricted; full-
reference requires original clean signal, which is not always available.

• Use case: Onsite benchmarking, codec evaluation, network tuning.

2. Glass-Box Parameter-Based Models

• How they work: Use detailed network and device parameters (e.g., packet loss,
delay, jitter) to predict ǪoE.

• Data from: Transport/network layers, signaling messages.

• Accuracy: Less precise than black-box but easier to deploy, especially online.

• Example: E-Model (ITU-T Rec. G.101) for voice ǪoE.

• Use case: Network planning and real-time service monitoring.

3. Gray-Box Models
• How they work: Combine black-box and glass-box approaches; use some
media info plus network/device parameters.

• Advantages: Better accuracy than glass-box; easier to deploy than black-box;

considers content characteristics.

• Use case: Increasingly popular for real-time, content-aware ǪoE measurement.

Tips for Selecting a ǪoS/ǪoE Mapping Model

• What operations are needed?

• What input data is available (signals, headers, payload)?

• Are there specification or usage constraints?

• Required precision level?

• Do you have all required inputs for the model?

8.3 IP-Oriented Parameter-Based ǪoS/ǪoE Mapping

Models
• Focused on measuring ǪoE for IP networks and services (like multimedia
streaming).

• IP networks carry media as packets, allowing runtime access to network and

application parameters.

• ǪoE varies with time in IP networks, unlike more static telecom networks.

8.3.1 Network Layer Models for Video Services

• Use network parameters only (e.g., packet loss, jitter, delay, bandwidth).

• Example model by Ketyko et al. uses audio/video packet loss and jitter plus
signal strength.

• Kim and Choi proposed a two-stage model combining parameters into a ǪoS
metric, then mapping that to ǪoE.
8.3.2 Application Layer Models for Video Services
• Use application layer parameters and user behavior.

• Example parameters:

o Start-up latency (waiting time before video plays)

o Number of quality switches during playback

o Number and duration of rebuffering events

• Khan et al. model accounts for frame rate, bit rate, and packet errors.

• Models may also differentiate video content types (slow, gentle, rapid
 movement).

• Kuipers et al. model considers startup latency and channel switching (zapping)

8.4 ǪoE vs ǪoS Service Monitoring

Monitoring is vital for IT systems to measure performance, detect faults, and guide
corrective actions.

8.4.1 Types of Monitoring:

• Network Monitoring: Measures path/link performance (throughput, loss, delay, jitter,

etc.).

• Infrastructure Monitoring: Measures device performance (CPU, memory, IO load).

• Platform Monitoring: Measures data center or virtualized infrastructure

performance.

• Service Monitoring: Measures service/application performance (technical C

perceptual metrics).

Monitoring systems use distributed probes and a central manager communicating

usually via UDP.

8.4.2 ǪoS Monitoring Solutions:

• Often deployed in data centers/clouds.

• Example: IPTV cloud service monitored via virtual probes (Vprobes) that track virtual
machine states and video packet flows.

• Data is collected into session-level detailed records (SDRs).

8.4.3 ǪoE Monitoring Solutions:

• Depend heavily on ǪoE/ǪoS mapping models.

• Four operational modes based on measurement and model location (Network (N),
Client (C), Both (B)):
G.1 Basic Concepts of Cloud Computing

• Cloud computing means providing IT resources (like servers,

storage, applications) over the Internet on demand.

• Defined by NIST as: Easy, on-demand access to a shared pool of

configurable computing resources that can be quickly provided or
released with little management effort.

• Benefits:

o Cost savings by paying only for what is used.

o No need for users/companies to maintain hardware or software.
o Professional network and security management.
• Cloud networking supports cloud services by connecting users and
cloud resources, possibly bypassing the Internet via private
networks for better performance and security.

• Cloud storage is a part of cloud computing, letting users store and manage
data remotely without owning physical storage hardware.

G.2 Cloud Service Models (NIST Defines 3 Main Types)

G.2.1 Software as a Service (SaaS)

• Provides software applications over the Internet.

• No need to install or maintain software locally.

• Accessible via web browsers or apps.

• Examples: Google Gmail, Microsoft 365, Salesforce, Cisco WebEx.

• Common users: companies and individuals who want easy

access to applications.

• Popular SaaS service categories include:

G.2.2 Platform as a Service (PaaS)
• Provides a platform for developers to build, deploy, and manage applications.

• Includes tools like programming languages, runtimes, databases, testing

environments.

• Useful for organizations developing custom apps without managing hardware.

• Examples: Google AppEngine, Microsoft Azure, Heroku.

• Common PaaS services include:

o Big Data, Business Intelligence, Databases, Development/Testing

platforms, General Purpose Platforms, Integration services.

G.2.3 Infrastructure as a Service (IaaS)

• Provides virtualized computing resources like servers, storage, and networking.

• Customers can provision and control resources themselves.

• Examples: Amazon EC2, Google Compute Engine, Microsoft Azure.

• Common IaaS services include:

o Backup/Recovery, Cloud Brokers (manage multiple clouds), Compute

resources, Content Delivery Networks (CDN), Service Management,
Storage.

G.2.4 Other Cloud Services (As per ITU-T Y.3500)

• Communications as a Service (CaaS): Video conferencing, messaging, VoIP.

• Compute as a Service (CompaaS): Simplified compute resource provisioning.

• Data Storage as a Service (DSaaS): Leasing storage space with management tools.

• Network as a Service (NaaS): Network services like VPN, bandwidth on

demand, security.

• Other emerging categories:

o Database as a Service, Desktop as a Service, Email as a Service, Identity

as a Service, Management as a Service, Security as a Service.
10.1 COMPONENTS OF IOT-ENABLED THINGS
Key Components:

1. Sensors:

o Measure physical, chemical, or biological parameters and convert them into

electronic signals (analog or digital).

o May send data actively (periodically or when threshold is crossed) or

passively (when requested).

2. Actuators:

o Convert electrical signals into physical, chemical, or biological actions.

o Types include:

▪ Hydraulic (use fluid power for motion)

▪ Pneumatic (use compressed gas)

▪ Electric (powered by motors)

▪ Mechanical (convert rotary to linear motion using gears, pulleys, etc.)

3. Microcontrollers:

o Provide embedded intelligence ("smart" capability).

o Embedded systems are computers designed for specific tasks, not

general-purpose.

o Microcontroller chip includes processor, memory (ROM/flash, RAM),

clock, and I/O control unit.

4. Transceivers:

o Devices that can both transmit and receive data.

o Most IoT devices use wireless transceivers (Wi-Fi, ZigBee, etc.).

5. RFID (Radio-Frequency Identification):

o Uses electronic tags to identify and track items remotely.

o Tags have a chip and antenna; readers detect tags and send info to
computer systems.

o Applications: tracking items/animals/people, payments, access control, anti-

counterfeiting