Part A – Answer All Questions (2 Marks

each)

[Link] is the Hype Cycle relevant to IoT and IIoT

technologies?
The Gartner Hype Cycle shows the stages of emerging
technology adoption—from innovation trigger to maturity.
In IoT / IIoT, it helps industries identify realistic expectations,
avoid hype, and decide when to invest in stable and proven
IoT technologies.

2. What is edge analytics?

Edge analytics refers to processing and analyzing data near
the source of generation (edge devices or gateways) instead
of sending it all to the cloud.
It reduces latency, bandwidth usage, and improves real-time
decision-making.

3. What is a cloud broker?

A cloud broker is an intermediary that manages, negotiates,
and integrates services from multiple cloud providers.
It helps users choose optimal cloud resources, ensures
interoperability, and monitors performance.

4. Distinguish between a sensor and a transducer.

  Sensor: Detects physical quantities (like temperature or
pressure) and converts them into an electrical signal.
 Transducer: A broader term—converts one form of
energy into another (a sensor is a type of transducer).

5. What is the frequency superimposed on the 4–20 mA

signal in HART communication?
HART (Highway Addressable Remote Transducer)
superimposes a 1,200 Hz and 2,200 Hz FSK (Frequency Shift
Keying) signal on the 4–20 mA analog loop to carry digital
data.

6. Define DoS attack in the context of network security.

A Denial of Service (DoS) attack attempts to make a network
or system unavailable to legitimate users by overwhelming it
with excessive traffic or resource requests.

7. State a few management aspects of cyber security.

1. Risk assessment and mitigation
2. Security policy formulation
3. Access control and authentication
4. Continuous monitoring and incident response

8. Spell out the data visualization tools used in IoT analytics.

Common tools: Power BI, Tableau, Grafana, Kibana, Google
Data Studio, and Matplotlib / Plotly for real-time
visualization of IoT data streams.

9. Why is correlation analysis useful in IoT data analytics for

predictive maintenance?
Correlation analysis identifies relationships between sensor
parameters (e.g., temperature vs vibration).
Detecting such dependencies helps predict failures early and
schedule maintenance proactively.

10. Define IIRA.

IIRA – Industrial Internet Reference Architecture provides a
standardized architectural framework for developing IIoT
systems.
It defines viewpoints—business, usage, functional, and
implementation—for scalable and secure industrial IoT
deployments.

PART B:

11 (a) Draw and explain the IIoT system architecture. Discuss

how each layer contributes to data acquisition, processing,
and decision-making.
Industrial IoT (IIoT) System Architecture:
IIoT architecture is generally divided into five layers:
1. Perception Layer (Sensing Layer):
o Contains sensors, actuators, RFID tags, etc.
o Collects physical parameters such as temperature,
vibration, flow, pressure, etc.
o Converts physical data into digital signals.
o Example: A temperature sensor in a machine
transmits real-time data.
2. Network Layer:
o Transfers the sensed data to the processing center.
o Uses communication protocols like Wi-Fi, Ethernet,
6LoWPAN, ZigBee, LoRaWAN, and Cellular IoT.
o Ensures secure and reliable data transmission.
3. Edge Layer (Edge Computing / Gateway):
o Performs local processing and analytics to reduce
latency.
o Aggregates and filters raw data before sending to
the cloud.
o Example: Gateway devices performing real-time
analytics near machines.
4. Application Layer (Cloud / Platform Layer):
 o Performs data storage, advanced analytics,
visualization, and decision-making.
o Uses AI/ML algorithms to extract insights and
predict failures.
o Examples: AWS IoT, Azure IoT Hub, ThingSpeak,
Google IoT Core.
5. Business Layer:
o Defines goals, user interfaces, and business logic.
o Generates actionable insights and reports for
management decision-making.
Data Flow:
Sensor → Gateway → Network → Cloud → User Dashboard
Benefits:
 Real-time monitoring
 Predictive maintenance
 Increased productivity
 Reduced downtime.

11(b) Short Notes

(i) Cost vs Ease of Prototyping (7 Marks)

Introduction
Prototyping in IoT and IIoT refers to the process of creating a
working model of a concept before it is implemented on a
large scale. It allows developers and engineers to test the
functionality, performance, and feasibility of their ideas using
hardware and software tools.
In the industrial IoT environment, cost and ease of
prototyping are two major factors that directly influence the
speed of development, innovation, and scalability of a
project.

1. Cost of Prototyping
Cost includes expenses related to:
 Hardware components (sensors, actuators,
microcontrollers)
 Software licenses and development tools
 Cloud connectivity and data storage
 Power consumption and maintenance
Low-cost prototyping has become possible through the
availability of open-source platforms like:
 Arduino, ESP32, Raspberry Pi
 NodeMCU, Particle Photon, and BeagleBone
These affordable devices help industries and students build
IoT solutions at minimal cost, encouraging experimentation
and innovation.
However, low-cost prototypes may have limitations, such as:
 Less accuracy and durability
 Limited processing power
 Inability to operate in harsh industrial conditions

2. Ease of Prototyping
Ease of prototyping refers to how quickly and efficiently an
IoT solution can be developed, tested, and modified.
Modern IoT ecosystems provide a variety of tools and
platforms that simplify prototyping:
 Cloud Platforms: AWS IoT, Azure IoT Hub, ThingSpeak
 Simulation Tools: Cisco Packet Tracer, Proteus, MATLAB
IoT Simulator
 Plug-and-play components: Modular sensors and
shields for Arduino and Raspberry Pi
These simplify connection, coding, and data visualization—
making it easy even for non-experts to create working
models.

3. Balancing Cost and Ease

Choosing the right combination of low cost and high ease is
crucial.
For example:
  Educational and research prototypes can use low-cost
microcontrollers.
 Industrial prototypes may use high-end boards (like
Intel NUC, NI cRIO) for durability and reliability.
Trade-off:
 Low-cost → fast experimentation but less reliability.
 High-cost → better stability but slower innovation.

4. Example
A startup developing a smart irrigation system may begin
with Arduino UNO and DHT11 sensors for quick testing (low
cost, high ease).
Once successful, it can shift to industrial-grade controllers
and weatherproof sensors for deployment (higher cost,
higher reliability).

(ii) Predictive Maintenance (6 Marks)

Introduction
Predictive maintenance (PdM) is one of the most valuable
Industrial IoT (IIoT) applications.
It uses real-time sensor data and data analytics to predict
when equipment will fail, allowing maintenance before actual
breakdowns occur.
1. Working Principle
1. DataAcquisition:
IoT sensors (temperature, vibration, current, pressure)
continuously monitor machine conditions.
2. DataTransmission:
Data is sent via industrial networks (MQTT, Modbus,
OPC-UA) to gateways and cloud servers.
3. DataAnalysis:
Algorithms like regression, correlation, and machine
learning detect abnormal patterns or trends.
4. DecisionAction:
Predictive alerts trigger maintenance scheduling and
spare-part planning.

2. Technologies Used
 Sensors: Vibration (ADXL345), Temperature (DHT22),
Pressure (MPX5010)
 Communication Protocols: HART, Modbus, MQTT
 Cloud Analytics: AWS IoT Analytics, Azure ML Studio,
ThingSpeak
 AI Algorithms: Linear Regression, SVM, Decision Trees

3. Advantages
 Prevents unexpected machine breakdowns.
  Extends equipment life.
 Reduces maintenance costs and downtime.
 Increases productivity and safety.
 Provides data-driven insights for process optimization.

4. Example
In an automotive manufacturing plant:
 Vibration sensors attached to motor bearings monitor
oscillations.
 Sudden increase in vibration amplitude indicates wear
and tear.
 The system triggers a maintenance alert before
complete failure.
This saves cost, time, and prevents production losses.

5. Implementation Architecture
Layer Function
Collects physical data (vibration,
Sensor Layer
temperature).
Gateway Layer Preprocesses data and sends to the cloud.
Cloud Layer Stores, analyzes, and visualizes the data.
Layer Function
Application
Displays alerts and maintenance schedules.
Layer

6. Use Cases
 Predicting conveyor belt wear.
 Monitoring turbine vibration.
 Predicting HVAC system failures.
 Oil level and temperature monitoring in machinery.
---------------------------------------------------------------------------

13(a) Design a sensor-based IIoT system for temperature and

vibration monitoring in a factory setup?
Components :
 Sensors
o Temperature: PT100/NTC or digital sensors like
DS18B20 (industrial accuracy ±0.5°C) — more
reliable than DHT11 for industrial use.
o Vibration: 3-axis MEMS accelerometer (e.g.,
ADXL345 / MPU6050 / industrial-grade ICP
accelerometer for high accuracy).
 Node Microcontroller / Data Collector
 o ESP32 or Arduino MKR series for low-power
wireless nodes; ADC + SPI/I²C for connecting
sensors.
o Local timestamping (RTC) and buffering in case of
connectivity loss.
 Gateway / Edge Computer
o Raspberry Pi or industrial edge gateway (for harsh
environments) that aggregates multiple nodes,
performs edge analytics, and bridges to cloud.
 Communication
o Field level: Wired RS485 (Modbus) or wireless
(LoRa/LoRaWAN, Wi-Fi) depending on distance and
reliability needs.
o Upstream: MQTT over TLS to cloud / broker.
 Cloud Platform
o AWS IoT / Azure IoT / ThingSpeak for ingestion,
long-term storage, model training, and large-scale
analytics.
 Visualization
o Grafana / Power BI / Kibana dashboards for KPIs,
trends, alerts, and historical comparisons.
 Alerting / Integration
 o SMS / Email / Push notifications and integration
with CMMS (Computerized Maintenance
Management System).

3. Data flow & pre-processing

1. Acquisition: Sensors sample at required rates (e.g.,
vibration at 1–5 kHz for detailed frequency analysis or
100–500 Hz for envelope/vibration RMS; temperature at
1 sample/min or 1 sample/5s depending on dynamics).
2. Local filtering: Anti-aliasing filter + moving-average or
band-pass filters for vibration; calibration corrections for
temperature.
3. Feature extraction at node or gateway: compute
features like RMS, Kurtosis, Peak-to-Peak, FFT bins, trend
slopes, temperature derivative, rolling mean/std.
4. Event reduction / compression: transmit only features
or events (anomaly flags) frequently and raw high-rate
windows only when anomaly detected — saves
bandwidth.
5. Message format: JSON payload over MQTT, e.g.
{ "device":"motor01","ts":"2025-11-
11T08:00:00Z","temp_c":72.4,"vib_rms":1.23,"fft_peak_freq"
:3120,"anomaly":false }

4. Edge analytics & algorithms

  Threshold-based rules: immediate alerts for
temperature > threshold or vibration RMS > limit.
 Statistical monitoring: rolling z-score, EWMA to detect
gradual drift.
 ML models on edge/cloud: regression for remaining
useful life (RUL), classification (normal vs faulty), and
unsupervised models (Isolation Forest, autoencoders)
for anomaly detection.
 Example pipeline: Node extracts RMS + spectral features
→ gateway runs unsupervised model → if anomaly
detected send raw waveform + features to cloud for
deeper analysis and model retraining.

5. Alerting & workflows

 Severity levels: info (warning), critical (stop machine),
maintenance required.
 Actions: Automated ticket creation in CMMS, SMS to
technician, dashboard highlight, PLC interlock if critical.
 SLA considerations: define max notification latency (e.g.,
critical alerts < 30s).

6. Security & reliability

 Secure transport: MQTT over TLS; mutual authentication
using X.509 certificates.
  Device identity & access control: unique device
certificates, regular rotation.
 Fault tolerance: local buffering, store-and-forward on
gateway during network outage.
 Data integrity: CRC on sensor frames, timestamps,
sequence numbers.

7. Deployment considerations & practical tips

 Sensor placement: vibration sensors mounted on
bearing housings; temperature sensors near hotspots —
follow ASTM/ISO placement guidelines.
 Sampling choices: choose sampling frequency to
capture the frequency range of interest (bearing faults
often at kHz, gear mesh higher).
 Calibration & maintenance: scheduled recalibration,
sensor health checks (self-test).
 Environmental protection: IP-rated enclosures,
appropriate isolations for EMI in heavy electrical
environments.
 Scalability: design MQTT topics and data schemas for
horizontal scale (topic per plant/line/device).

8. Example: Minimal MQTT topic layout

 factory1/lineA/motor01/telemetry — periodic features
  factory1/lineA/motor01/event — anomaly / alarm
events
 factory1/lineA/motor01/raw — raw high-rate waveform
(only on request/anomaly)

9. KPIs & benefits

 KPIs to monitor: Mean Time Between Failures (MTBF),
Unplanned Downtime reduction, Alert precision & recall,
maintenance cost savings.
 Advantages: earlier fault detection, optimized
maintenance scheduling, reduced downtime, energy
savings, centralized asset visibility.

13 (b) Design and illustrate a MODBUS-based sensor network

for an industrial automation system. Explain the data frame
format and working of MODBUS-RTU?
. Introduction
In industrial automation, MODBUS is one of the most widely
used communication protocols for connecting sensors,
actuators, controllers, and PLCs in a reliable, standardized
way.
It allows devices to exchange information over serial lines
(RS-232 or RS-485) or Ethernet.
The most common implementation for serial communication
is MODBUS-RTU (Remote Terminal Unit).

2. Why MODBUS in Industrial IoT?

 Simple and open protocol.
 Supports master–slave / client–server architecture.
 Ensures deterministic and reliable communication.
 Compatible with microcontrollers, PLCs, RTUs, and smart
sensors.
 Cost-effective for industrial networks.

3. System Design and Architecture

(a) Components
1. Master Device (Controller / PLC):
Initiates communication by sending commands or
queries to slave devices.
Example: Siemens S7, Delta PLC, or Raspberry Pi acting
as master.
2. Slave Devices (Sensors / Actuators):
Respond to master requests.
Examples:
o Temperature sensor node
o Pressure transducer
 o Flow sensor
o Motor controller
3. Communication Medium:
o RS-485 (preferred): differential signaling for long-
distance, multi-drop networks (up to 1200 meters).
o RS-232: for short point-to-point communication.
4. SCADA / HMI System:
Displays real-time process values and status of sensors
via graphical interface.

(b) Network Topology (Diagram Description)

You can draw the following simple diagram in your notebook:
  The PLC (Master) communicates with all sensors (Slaves)
using RS-485 two-wire twisted pair bus.
 Each slave has a unique address (1–247).
 SCADA system connected to the PLC visualizes the data.

4. MODBUS-RTU Data Frame Format

Each MODBUS-RTU message frame consists of 4 major parts:
Size
Field Description
(bytes)
3.5 character times of silence
Start / Silent Interval -
marks frame boundary
Slave Address 1 Identifies the slave (1–247)
Defines the requested action
Function Code 1
(e.g., Read, Write)
Contains register address,
Data Field N
data value, etc.
CRC (Cyclic
2 Error checking code
Redundancy Check)

Example: Read Holding Register (Function Code = 03)

Field Value Description
Slave Address 01 Target device
Function Code 03 Read holding registers
Field Value Description

Starting Address 006B Register to start reading

No. of Registers 0003 Number of registers
CRC xxxx 2-byte error check
Frame: 01 03 00 6B 00 03 CRC
Response (from slave):
01 03 06 02 2B 00 00 00 64 CRC

5. Working of MODBUS-RTU Communication

1. Request Phase:
o The master device (PLC) sends a request to a
specific slave using its address and function code.
o Example: “Slave 1, send me your temperature
data.”
2. Response Phase:
o The addressed slave processes the request and
sends the corresponding data back to the master.
o The response includes data bytes and CRC for
verification.
3. Error Detection:
o If CRC does not match or slave address is incorrect,
the master discards the frame and may retry.
 o Error codes are returned to indicate issues like
“illegal function” or “invalid address.”
4. Timing:
o A minimum 3.5 character time gap marks the start
and end of a frame.
o This prevents overlapping of data frames in multi-
device networks.

6. Common MODBUS Function Codes

Code Function Description
01 Read Coils Reads ON/OFF status of coils
02 Read Discrete Inputs Reads digital inputs
03 Read Holding Registers Reads analog/digital values
04 Read Input Registers Reads sensor data
05 Write Single Coil Turns ON/OFF a single coil
06 Write Single Register Writes one data register
15 Write Multiple Coils Writes several outputs
16 Write Multiple Registers Writes multiple registers

7. Advantages of MODBUS-RTU in Industrial Automation

 Simplicity: Easy to implement and configure.
 Interoperability: Open and vendor-independent.
  Long-distance communication: Up to 1200 meters on
RS-485.
 Multi-device support: Up to 247 slaves.
 Low bandwidth usage: Efficient binary communication.
 Error checking: Reliable CRC ensures data integrity.

8. Applications
 Temperature, vibration, and pressure monitoring.
 Industrial process control.
 Energy metering and data logging.
 Building management systems (HVAC, lighting).

14 (a) Analyze the three cloud service models SaaS, PaaS, and
IaaS with respect to their architecture, functionalities, and
applications.
Introduction
Cloud computing provides on-demand access to computing
resources such as servers, storage, applications, and
networking through the Internet.
It eliminates the need for owning physical infrastructure and
enables scalability, flexibility, and cost-efficiency.
Cloud services are commonly classified into three main
models:
  Infrastructure as a Service (IaaS)
 Platform as a Service (PaaS)
 Software as a Service (SaaS)
Each model provides different levels of control, flexibility, and
management.

2. Overall Cloud Architecture

The cloud architecture generally has three layers:
1. Infrastructure Layer: Virtualized hardware resources —
servers, networks, and storage (IaaS).
2. Platform Layer: Runtime environments, databases, and
middleware for app development (PaaS).
3. Application Layer: Ready-to-use software accessible via
web or mobile interfaces (SaaS).
┌────────────────────────────┐
│ Software (SaaS) │ → End-user access (apps like Gmail,
Zoom)
├────────────────────────────┤
│ Platform (PaaS) │ → Developer tools, runtime, DBs
├────────────────────────────┤
│ Infrastructure (IaaS) │ → Virtual servers, networking,
storage
└────────────────────────────┘
3. Infrastructure as a Service (IaaS)
(a) Definition
IaaS provides virtualized computing infrastructure over the
internet.
It offers raw resources like virtual machines (VMs), networks,
and storage, which users can configure as needed.
(b) Architecture
 Physical Data Centers: Host physical servers and storage.
 Virtualization Layer: Hypervisors (like VMware, Hyper-V)
create virtual resources.
 Resource Management Layer: Allocates and monitors
resources.
 User Interface/API Layer: Allows users to create and
manage VMs and networks.
(c) Functionalities
 Provisioning of servers and storage on demand.
 Scalable resource allocation.
 Network configuration and load balancing.
 Data backup and disaster recovery.
 Security and access control.
(d) Examples
 Amazon Web Services (EC2, S3)
  Microsoft Azure Virtual Machines
 Google Compute Engine
(e) Applications
 Hosting websites or enterprise servers.
 Big data analysis and virtual networks.
 Backup and disaster recovery solutions.
 Creating private or hybrid cloud environments.

4. Platform as a Service (PaaS)

(a) Definition
PaaS provides a complete development and deployment
environment in the cloud.
It offers tools, frameworks, and runtime environments to
build, test, and deploy applications without managing
infrastructure.
(b) Architecture
 Infrastructure Layer: Managed by the provider (IaaS
base).
 Runtime and Middleware: Includes OS, web servers,
databases, and frameworks.
 Developer Tools: APIs, SDKs, version control, and testing
tools.
  Application Layer: End-user applications built by
developers.
(c) Functionalities
 Application development and hosting.
 Database management and middleware integration.
 Auto-scaling and load balancing.
 Continuous integration and deployment (CI/CD).
 Support for multiple programming languages.
(d) Examples
 Google App Engine
 Microsoft Azure App Services
 Heroku
 IBM Cloud Foundry
(e) Applications
 Rapid development of web and mobile applications.
 Building APIs and microservices.
 IoT data analytics platforms.
 Educational and research development environments.

5. Software as a Service (SaaS)

(a) Definition
SaaS provides ready-to-use software applications over the
Internet.
Users can access these apps via browsers or mobile devices
without worrying about installation, updates, or
maintenance.
(b) Architecture
 Multi-tenant Architecture: One application instance
serves multiple customers.
 Data Layer: Stores user data securely in the cloud.
 Application Layer: Handles user interface and business
logic.
 Access Layer: Provides web or API-based access.
(c) Functionalities
 Centralized hosting and management.
 Subscription or pay-per-use model.
 Regular updates and automatic scaling.
 Data security and backup by the provider.
 Accessible from any device with Internet.
(d) Examples
 Google Workspace (Docs, Drive, Gmail)
 Salesforce CRM
 Zoom, Slack, Dropbox
  Microsoft 365
(e) Applications
 Office productivity tools.
 Customer relationship management (CRM).
 Enterprise resource planning (ERP).
 Communication and collaboration tools.
 Online learning platforms.

6. Comparison Table
Feature IaaS PaaS SaaS
Minimal
Full control Control over control;
User Control over applications provider
infrastructure only manages
everything
Managed_By Hardware & Hardware,
Complete stack
Provider virtualization OS, runtime
IT
Target Users Developers End users
administrators
Customization High Medium Low
Subscription
Pay per use Subscription
Cost Model for dev
(VMs, storage) per user/app
platform
Feature IaaS PaaS SaaS
AWS EC2, Azure Google App Gmail,
Example
VMs Engine Salesforce

9. Relationship Among SaaS, PaaS, and IaaS

Each layer builds upon the previous one:
 IaaS provides base infrastructure →
 PaaS builds development environment on top of it →
 SaaS offers complete user applications using PaaS and
IaaS services underneath.
Thus, SaaS = PaaS + IaaS (in functional dependence).

8. Advantages of Cloud Service Models

1. Cost Efficiency: Pay-as-you-go pricing reduces hardware
investment.
2. Scalability: Resources scale automatically based on
demand.
3. Accessibility: Services available globally over the
Internet.
4. Maintenance-Free: Software updates handled by
providers.
5. Reliability: High availability and backup features.
 6. Security: Data protection and access management
systems.

9. Real-Time Industrial Applications

Industry Use of Cloud Models
IaaS for machine data storage; PaaS for
Manufacturing
analytics; SaaS for dashboards
SaaS (Google Classroom), PaaS for app
Education
development
SaaS for patient management; IaaS for
Healthcare
medical imaging data
IaaS for data hosting, PaaS for IoT app
IoT Systems
deployment, SaaS for monitoring dashboards

14 (b) Layered Attacker Model in Network Security?

Introduction
In modern networked systems, threats come from multiple
directions — users, insiders, external hackers, and even
software bugs.
To understand, prevent, and defend against these attacks,
security analysts use the Layered Attacker Model.
This model explains how attacks occur at different layers of a
network system and helps design multi-layered security
strategies for protection.
2. Concept of Layered Security
Layered security (also called Defense in Depth) means
applying multiple levels of protection — just like multiple
locks on a door.
If one layer fails, others continue to protect the system.
Each layer (Physical, Network, Application, etc.) has different
attacker types, vulnerabilities, and defenses.

3. Need for Layered Attacker Model

 Network attacks can target any level of the system.
 A single firewall or antivirus cannot protect the entire
system.
 Layered modeling helps identify vulnerabilities in every
layer.
 Helps security engineers design appropriate
countermeasures.
 Essential for cyber risk assessment and incident
response planning.

4. Layers in the Attacker Model

The Layered Attacker Model corresponds roughly to the OSI
Model — each network layer is vulnerable to specific attack
types.
Let’s analyze each layer in detail .

(i) Physical Layer Attacks

Role: Transmits raw bits over physical media like cables or
wireless links.
Possible Attacks:
 Wire tapping: Unauthorized interception of data cables.
 Jamming: Disrupting wireless signals using radio
interference.
 Hardware tampering: Replacing or modifying network
devices.
 Theft or damage of equipment or routers.
Defenses:
 Secure physical access (locks, surveillance, biometric
systems).
 Shielded cables and signal encryption.
 Environmental monitoring and alarms.

(ii) Data Link Layer Attacks

Role: Responsible for reliable data transfer between adjacent
network nodes (Ethernet, Wi-Fi).
Possible Attacks:
  MAC spoofing: Attacker changes MAC address to
impersonate another device.
 ARP poisoning: Sends false ARP messages to redirect
traffic.
 VLAN hopping: Unauthorized access between VLANs.
 Wi-Fi eavesdropping using fake access points.
Defenses:
 MAC filtering and ARP inspection.
 VLAN configuration and port security.
 Use of WPA3/WEP encryption in wireless networks.

(iii) Network Layer Attacks

Role: Handles logical addressing and routing (IP layer).
Possible Attacks:
 IP spoofing: Attacker sends packets using fake IP
addresses.
 Routing attacks: Modify routing tables to misroute
traffic.
 DoS (Denial of Service): Overload network with junk
packets.
 Man-in-the-Middle (MITM): Intercept packets between
sender and receiver.
Defenses:
  Firewalls and Intrusion Detection Systems (IDS).
 Packet filtering and route authentication.
 IPsec (IP Security) for encrypted communication.

(iv) Transport Layer Attacks

Role: Ensures reliable end-to-end communication using
TCP/UDP.
Possible Attacks:
 TCP SYN Flood: Exhausts server connection queue.
 UDP Flooding: Sends large volumes of UDP packets to
overwhelm host.
 Session hijacking: Attacker takes over an active session.
Defenses:
 Rate limiting and SYN cookies.
 Secure session management and sequence number
randomization.
 Firewalls and deep packet inspection.

(v) Application Layer Attacks

Role: Supports end-user applications like web, email, FTP, etc.
Possible Attacks:
  SQL Injection: Inject malicious SQL queries into input
fields.
 Cross-Site Scripting (XSS): Injects scripts into webpages.
 Phishing: Trick users to reveal credentials.
 Malware and ransomware.
Defenses:
 Input validation and sanitation.
 Web Application Firewalls (WAF).
 Regular software patching and user awareness training.

5. Attacker Types in the Model

Attacker Type Description Example
Amateur hacker using Runs DDoS tools
Script Kiddie
prebuilt tools from Internet
Employee or partner Steals confidential
Insider Attacker
misusing access data
Motivated by Credit card theft,
Cybercriminal
financial gain ransomware
Political or social
Hacktivist Defaces websites
motive
State-sponsored Government-backed Attacks on critical
Attacker cyber espionage infrastructure
Each attacker targets different layers and assets depending
on motive and skill level.

6. Layered Defense Strategy

A strong security system applies multiple layers of defense,
known as Defense-in-Depth.
Security Layers and Mechanisms:
Layer Security Mechanism
Physical CCTV, biometric access, alarms
Network Firewalls, VPNs, IDS/IPS
Host Antivirus, OS hardening, patch management
Application Secure coding, authentication, encryption
Data Access control, database encryption
Human Layer Training, awareness, strong password policies
By combining these layers, even if one fails, others continue
to protect the system.

7. Example of Layered Attack and Defense

Scenario:
A hacker tries to steal sensor data from an IoT system in a
factory.
 Physical Layer: He gains access to Wi-Fi access point.
→ Defense: MAC filtering and CCTV detection.
  Network Layer: Tries IP spoofing to impersonate
devices.
→ Defense: IPsec encryption and authentication.
 Application Layer: Tries to inject malicious script in
dashboard.
→ Defense: Input validation and WAF.
Hence, each layer resists the attack, ensuring system
integrity.

8. Advantages of Layered Attacker Model

1. Helps identify vulnerabilities in every OSI layer.
2. Provides a systematic approach to design secure
networks.
3. Improves incident detection and response time.
4. Enhances resilience against multiple attack types.
5. Supports risk analysis and security audits.
6. Reduces the possibility of total system compromise.

9. Limitations
 Increased complexity and cost.
 Requires continuous maintenance and updates.
 Misconfiguration in one layer may weaken others.
15 (a) Evaluate with an example how regression analysis can
be implemented to predict crop yield in a smart agriculture
IoT system by assuming relevant sensor data such as soil
moisture, temperature, and humidity.?
Introduction
Smart Agriculture is one of the most important applications
of the Internet of Things (IoT).
IoT sensors deployed in farms continuously collect
environmental and soil parameters such as:
 Soil moisture
 Temperature
 Humidity
 Rainfall, pH, and sunlight intensity
Analyzing this sensor data helps farmers take data-driven
decisions to improve productivity and crop health.
One of the powerful data analysis techniques used in IoT-
based agriculture is Regression Analysis — a statistical
method used to model the relationship between dependent
and independent variables to predict future outcomes.

2. Definition of Regression Analysis

Regression Analysis is a supervised machine learning or
statistical technique that finds the relationship between one
dependent variable (target) and one or more independent
variables (predictors).

3. Need for Regression in Smart Agriculture

IoT sensors generate continuous, real-time data that can be
analyzed to find patterns affecting crop yield.
Regression helps to:
 Predict crop yield before harvest.
 Optimize irrigation and fertilizer usage.
 Understand environmental impacts on plant growth.
 Plan harvesting schedules based on forecasted
productivity.
Thus, regression analysis enables precision farming using
data-driven insights.

5. Implementation of Regression Analysis

Step 1: Data Collection
IoT sensors collect continuous real-time data such as:
Parameter Sensor Used Example Data
Soil Moisture (%) Capacitive Sensor 40, 45, 50
Parameter Sensor Used Example Data
Temperature (°C) DHT11 Sensor 25, 28, 30
Humidity (%) DHT11 Sensor 60, 65, 70
Crop Yield (kg/acre) Manual Input 1800, 2000, 2200

Step 2: Data Preprocessing

 Remove noise and missing values.
 Normalize the sensor data.
 Store in a cloud database (e.g., AWS IoT, ThingSpeak, or
Azure IoT Hub).

Step 3: Model Training

 Historical IoT sensor data from multiple crop cycles is fed
into the model.
 Using algorithms like Least Squares Method, model
 The model learns how each sensor parameter affects
crop yield.

Step 5: Prediction
Once trained, the model can predict yield for new sensor
inputs.
6. Example Visualization
IoT dashboard displays:
 Real-time graphs of soil and climate data.
 Regression trend line showing relation between
variables and yield.
 Alerts if parameters move outside ideal ranges.
(You can sketch a simple graph of yield vs soil moisture with a
rising linear trend line in exams.)

7. Evaluation and Insights

Regression analysis helps in:
1. Identifying Key Factors:
Understand which parameters (e.g., soil moisture) most
influence yield.
2. ResourceOptimization:
Farmers can adjust irrigation and fertilizers based on
predicted yield.
3. DecisionSupport:
Early yield forecasts help in market planning and
logistics.
4. Scalability:
The same model can be deployed across multiple farms
using cloud IoT platforms.
8. Extensions of Regression Models in IoT Agriculture
Type of Regression Application
Basic yield prediction using few
Linear Regression
parameters
Polynomial For nonlinear relations between
Regression weather and yield
Classify crops as “high yield” or “low
Logistic Regression
yield”
Time Series
Forecast yield over future seasons
Regression

9. Advantages
 Simple to implement and interpret.
 Works well with continuous IoT sensor data.
 Provides accurate, quantitative predictions.
 Supports automation and smart farming decisions.

10. Challenges
 Sensor data may contain noise or missing values.
 Environmental factors like pests or rainfall are hard to
quantify.
 Regression assumes linear relationships — not always
true.
  Requires large datasets for high accuracy.

15 (b) Using Bosch’s example, assess how correlation and

regression techniques support data-driven decision-making in
IIoT-enabled production lines.

Introduction
In the era of Industry 4.0, industries are becoming
increasingly automated and data-driven through the use of
Industrial Internet of Things (IIoT).
IIoT enables machines, sensors, and control systems to
communicate and exchange real-time data for smart
decision-making.
One of the leading adopters of IIoT is Bosch, a global
manufacturer known for integrating advanced data analytics
into their production systems.
Two of the most widely used analytical techniques in IIoT
environments are:
 Correlation Analysis – To identify relationships between
process variables.
 Regression Analysis – To build predictive models for
optimizing production efficiency.

2. Bosch IIoT Production System Overview

Bosch uses smart sensors and cloud analytics to monitor and
control production processes in their factories.
Architecture:
1. Sensors_and_Devices:
Installed on machines to monitor temperature,
vibration, speed, and energy consumption.
2. Edge_Gateways:
Collect data from multiple machines and send it securely
to the cloud.
3. Bosch_IoT_Suite&Cloud_Platform:
Stores large-scale industrial data and runs analytics
models.
4. Analytics_Engine:
Applies correlation and regression algorithms to predict
faults, optimize performance, and reduce downtime.
5. Dashboard_Interface:
Displays insights for engineers to make data-driven
decisions.

3. Understanding Correlation and Regression

Correlation Analysis
 Measures the degree of relationship between two
variables.
  Represented by the Correlation Coefficient (r) ranging
from -1 to +1.
o r = +1: Perfect positive correlation
o r = -1: Perfect negative correlation
o r = 0: No relationship
Regression Analysis
 Builds a predictive model that estimates one variable
based on others.
 Example: Predict machine output (Y) based on vibration
and temperature (X₁, X₂).
Linear Regression model:

4. Application in Bosch’s Smart Factory

Bosch’s IIoT production lines use thousands of sensors to
monitor:
 Motor vibration levels
 Conveyor speed
 Tool temperature
 Power usage
 Cycle time
 Product quality metrics
All these parameters are collected in real time and analyzed
using correlation and regression to improve efficiency.

5. Step-by-Step Working Example

Step 1: Data Collection
Bosch installs temperature and vibration sensors on CNC
machines.
Step 2: Correlation Analysis
 Correlation between temperature and output = -0.89 →
strong negative correlation
 Correlation between vibration and output = -0.82 →
strong negative correlation
✅ This means as temperature and vibration increase, output
decreases.
Step 3: Regression Analysis
Bosch develops a multiple regression model:
6. Decision-Making Based on Analytics
Using these analytical results, Bosch engineers can take data-
driven actions, such as:
 Reducing spindle temperature using smart cooling.
 Scheduling preventive maintenance when vibration
crosses threshold.
  Adjusting machine speed to maintain optimal
production quality.
The correlation identifies which factors most affect
performance, while regression helps predict future outcomes
if conditions change.

7. Benefits of Using Correlation & Regression in IIoT

Benefit Explanation
Predictive Detects machine faults before breakdown
Maintenance using sensor patterns.
Predicts product defects based on sensor
Quality Control
variables.
Process Identifies optimal machine parameters for
Optimization best performance.
Correlation between power usage and
Energy Efficiency
machine load helps reduce waste.
Prevents downtime and unplanned
Cost Reduction
maintenance.
Cloud dashboards show performance
Real-time Insights
metrics instantly.

8. Bosch Case Example

In Bosch’s Homburg Smart Factory (Germany):
  Over 200 sensors collect 500+ parameters per minute.
 Machine-learning regression models predict valve
production yield.
 Correlation analysis between temperature, torque, and
cycle time identifies the main cause of defective valves.
 Result: 15% reduction in downtime and 5% increase in
production rate.
This real example shows how data analytics enhances
productivity and reliability.
[Link] in Implementation
1. Handling huge real-time data (Big Data issues).
2. Need for high-speed cloud and edge computing.
3. Noise and inaccuracy in sensor readings.
4. Continuous model retraining required for accuracy.

PART -C

16(a) — Designing a Wearable System for Elderly Fall

Detection & Emergency Response ?
Introduction (short)
A wearable fall-detection system for elderly patients must
reliably detect falls, minimise false alarms, and deliver rapid
emergency response while conserving battery and protecting
user data. The design combines sensors, local processing
(edge), communications, alerting mechanisms and security
provisions.

2. Design Objectives (what the system must do)

 Detect falls with high accuracy (high sensitivity) and low
false-alarm rate (high specificity).
 Respond quickly (notify caregiver/emergency services
within seconds).
 Preserve battery life for days of continuous operation.
 Ensure privacy & security of health and location data.
 Be comfortable and robust for daily use (IP rating,
wearability).

3. Functional specifications
1. Sensing & Detection
o Detect impact, orientation change, and lack of
motion after impact.
o Measure physiological context (optional): heart
rate, SpO₂ to confirm emergency.
2. On-device Processing
 o Real-time signal processing and classification (edge
AI) to decide fall vs non-fall.
3. Communication
o Short-range: BLE to smartphone gateway.
o Long-range: LTE-M / NB-IoT or Wi-Fi (direct device
cloud) as fallback.
o Local alerts: audible buzzer and vibration for user
acknowledgement.
4. Alerting & Response
o Local audible/vibration alert allowing the user to
cancel false alarm.
o If unconfirmed, automatic message with location
and sensor snapshot to caregiver and emergency
services.
o Two-way voice call capability (through smartphone
or built-in cellular) for triage.
5. User Interface
o Mobile app for caregivers: real-time telemetry, alert
history, location tracking, and settings.
o Web dashboard for hospital/clinic integration and
logging.
6. Power Management
 o Low-power hardware and configurable sampling to
meet target battery life (e.g., 5–7 days typical).
7. Security & Privacy
o End-to-end encryption, device authentication,
secure OTA updates, and privacy-preserving data
handling.

4. Technical specifications (example values suitable for

exam)
Hardware
 Form factor: Wristband / pendant.
 Microcontroller / SoC: ARM Cortex-M4/M33 with DSP +
hardware crypto.
 Sensors:
o 3-axis accelerometer (±16 g, e.g., 100–200 Hz
sampling) — for impact and orientation.
o 3-axis gyroscope (200–500 Hz) — for angular
velocity.
o Barometer (optional) — to detect height change.
o Heart-rate sensor (PPG) and SpO₂ sensor (optional)
— for medical context.
o Magnetometer (optional) — orientation correction.
  Communications: BLE 5.0 (primary), LTE-M/NB-IoT
module (optional), Wi-Fi (optional).
 Location: Assisted GPS via smartphone; fallback: cell-
tower triangulation.
 Battery: 400–600 mAh Li-ion, target 5 days with typical
use.
 User feedback: Vibration motor, buzzer, LED, one
physical help button.
 Enclosure: IP67, operating temperature range 0–50°C.
Software / Algorithms
 Sampling rates: Accelerometer 100–200 Hz; Gyroscope
200–500 Hz when high-resolution needed; otherwise
duty-cycle.
 Edge algorithm: Sensor fusion + lightweight classifier
(threshold + decision tree or tiny neural network /
autoencoder anomaly detector).
 False-alarm rejection: Post-impact inactivity check (e.g.,
>30 s no movement) + pulse/SpO₂ corroboration.
 Local buffer: Store pre/post-event raw window (e.g., 2 s
pre, 10 s post) for upload on alert.
Latency & Performance Targets
 Detection latency: < 2 s from fall event to local decision.
  Alerting latency: < 15 s from detection to caregiver
notification (dependent on network).
 Accuracy target: Sensitivity ≥ 95%, Specificity ≥ 90%
(after tuning).

5. System architecture & data flow (block diagram —

describe to draw)
Draw blocks: Sensors → MCU (edge processing) →
Communications module → Gateway (smartphone/cloud)
→ Cloud analytics & Dashboard → Caregiver / Emergency
services.
Include local alert (vibration/LED) and cancel button between
MCU and communications.
Flow: sensor readings → feature extraction (impact peak,
orientation change, RMS, jerk, tilt) → classifier → if fall then
trigger local alert → if not cancelled → send alert packet (ID,
timestamp, GPS, sensor snapshot) to cloud → cloud forwards
to caregivers / EMS / CMMS.

6. Justification & measures to address key concerns

A. Accuracy (reduce false positives/negatives)
 Sensor fusion: Combine accel + gyro + barometer + PPG
to corroborate fall. Falls show a characteristic sequence:
high acceleration peak, rapid orientation change, then
 inactivity. Fusion reduces misclassification (e.g., sitting
abruptly).
 Feature engineering: Use time-domain (peak
acceleration, impact duration) and frequency-domain
(FFT energy) features and statistics (RMS, kurtosis).
 Adaptive thresholds & personalization: Calibrate
thresholds per user (age, gait) and allow model
personalization on-device/cloud.
 Edge ML: Use a lightweight classifier trained on labeled
fall & ADL (activities of daily living) datasets. Retrain
periodically with anonymized aggregated data.
 Anti-tamper & self-test: Periodic calibration and sensor
health checks; reject data when sensors out-of-range.
 User confirmation window: After detection, vibrate and
allow user 30 s to cancel — reduces false alarms.
B. Battery life
 Duty-cycling & adaptive sampling: Use low sampling
(e.g., 10 Hz) normally; switch to high-rate (100–200 Hz)
when motion detected (wake-on-motion).
 Edge processing: Do classification locally to avoid
energy-costly continuous transmission. Transmit only on
events or periodic health telemetry.
 Low-power components: Use BLE 5.0, low-power MCU
with sleep modes, efficient sensors with FIFO buffers.
  Hardware optimization: Efficient power regulators,
battery fuel-gauge, and user-configurable reporting
intervals.
 Example calculation: If average consumption is 20 mA
active, 2 mA sleep; with 400 mAh battery and 10% active
duty = battery life ≈ 400 / (0.120 + 0.92) ≈ 400 / (2 + 1.8)
≈ 400 / 3.8 ≈ 105 h ≈ 4.4 days. Tuning duty cycle can
reach 5–7 days.
C. Data security & privacy
 Device identity & authentication: Unique device
certificates (X.509) or pre-shared keys; mutual TLS
between device/gateway/cloud.
 Encryption: TLS/DTLS for transport; AES-128/256 for
data at rest on cloud.
 Access control: Role-based access for caregivers; audit
logs for data access.
 Secure boot & OTA: Firmware signed and verified;
secure OTA updates to patch vulnerabilities.
 Anonymization & minimisation: Upload only necessary
data; store raw windows temporarily; use hashed IDs for
analytics; comply with data protection rules (e.g.,
consent, retention policies).
 Fail-safe: If network unavailable, store events encrypted
locally and transmit when online.
7. Emergency response workflow (operational sequence)
1. Fall detected → local alarm & vibration.
2. No cancel → package alert: patient ID, timestamp, last
GPS, sensor snapshot.
3. Send to cloud → notify primary caregiver (app push,
SMS) and fallback emergency contacts.
4. If no response → escalate to emergency services with
voice call option (via smartphone or cellular module).
5. Log incident in cloud and create ticket in CMMS or
hospital system.

8. Testing, validation & compliance

 Validation: Test using benchmark ADL and fall datasets;
field trials with volunteers.
 Metrics: Report confusion matrix, ROC, precision/recall,
mean time to notify (MTTN).
 Regulatory: Follow medical device guidelines if classified
(CE/FDA), and privacy laws (GDPR/HIPAA)

16(b) Design a basic smart meter block diagram and explain

energy calculation, threats, and security techniques
Smart Meter Block Diagram and Energy Calculation (7
Marks)
A smart meter is an advanced electronic device that records
electrical energy consumption in real time and automatically
sends the data to the utility provider for monitoring and
billing.
Block Diagram (describe in words for exam)
The smart meter consists of the following major blocks:
1. Voltage and Current Sensors – These measure the
supply voltage and current drawn by the load.
2. Signal Conditioning Unit – Filters and amplifies the
measured signals to remove noise.
3. Analog-to-Digital Converter (ADC) – Converts the analog
sensor signals into digital form.
4. Microcontroller or DSP – The brain of the system that
processes data, calculates energy consumption, and
performs communication functions.
5. Communication Module – Transfers energy readings to
the central system using Wi-Fi, GSM, ZigBee, NB-IoT, or
PLC.
6. Display Unit – Shows readings like voltage, current,
power, and energy consumed.
7. Memory Unit – Stores meter readings and tariff
information.
 8. Power Supply Unit – Provides regulated power to all
components.
9. Security Module – Protects stored data and ensures
encrypted communication.
Working
 The voltage and current sensors continuously sense the
electrical parameters.
 These signals are processed by the signal conditioning
circuit and converted into digital values.
 The microcontroller calculates power usage and total
energy consumed over time.
 The processed data is displayed to the user and
transmitted securely to the utility provider through the
communication module.
 The system can also send alerts in case of overloads,
tampering, or abnormal usage patterns.
Purpose and Advantages
 Enables accurate billing and remote reading without
human intervention.
 Supports real-time monitoring, load management, and
energy theft detection.
 Helps both consumers and suppliers improve energy
efficiency.
(ii) Common Threats in Smart Energy Systems and Security
Techniques (8 Marks)
Common Threats
1. Data Tampering: Attackers may modify readings to
reduce bills or mislead the system.
2. Eavesdropping: Unauthorized users might intercept
communication between meter and server.
3. Replay Attacks: Old valid data is resent to confuse the
billing system.
4. Man-in-the-Middle Attack: Data is intercepted and
altered during transmission.
5. Firmware Tampering: Malicious code may be installed in
the meter to change its function.
6. Physical Tampering: Attempts to bypass sensors or open
the meter casing.
7. Denial of Service (DoS): Flooding the communication
channel to stop data reporting.
8. Privacy Leakage: Consumption data can reveal user
habits or presence at home.

Encryption and Authentication Techniques

1. Data Encryption:
 o Smart meters use secure algorithms like AES or
similar to ensure that transmitted data cannot be
read by unauthorized parties.
o Encryption makes the information confidential
while traveling over networks.
2. Authentication:
o Both the meter and the utility server verify each
other’s identity before communication.
o Digital certificates or secure keys ensure that only
genuine devices can connect.
3. Message Integrity:
o Techniques like message signing or checksum
verification ensure that the data has not been
altered during transmission.
4. Secure Firmware Updates:
o Firmware is digitally signed by the manufacturer,
and only verified software can be installed.
5. Key Management:
o Unique secret keys are stored securely within each
meter to prevent cloning or misuse.
o Keys are rotated or updated periodically for better
protection.
6. Secure Communication Protocols:
 o Use of secure versions of MQTT, TLS, or DTLS
protocols ensures protected communication
between the meter and the cloud.
7. Tamper Detection:
o Physical tamper switches and magnetic sensors
detect unauthorized opening or bypassing
attempts.
8. Privacy Measures:
o Only necessary data is shared with utilities, and
personal information is kept encrypted.