1
MSc Dissertation
Lightweight Deep Learning–Based Intrusion Detection
for Resource-Constrained IoT Devices
Student Name: Arya Vishnu
Course: MSc Cyber Security and Forensic Information Technology
Academic Year: 2025–2026
Supervisor: Peter Bednar
Chapters 1–3 Draft
, 2
Table of Contents
CHAPTER 1: INTRODUCTION............................................................................................................2
1.1 Background............................................................................................................................3
1.2 Problem Statement................................................................................................................4
1.3 Research Aim and Objectives................................................................................................4
1.4 Research Question and Hypotheses......................................................................................5
1.5 Significance, Scope, and Limitations......................................................................................5
CHAPTER 2: LITERATURE REVIEW....................................................................................................6
2.0 Introduction...........................................................................................................................7
2.1 IoT Security Challenges..........................................................................................................7
2.2 Intrusion Detection System Types.........................................................................................8
2.3 Deep Learning in Intrusion Detection....................................................................................9
2.4 Lightweight Deep Learning Techniques...............................................................................10
2.5 Research Gap.......................................................................................................................10
CHAPTER 3: RESEARCH METHODOLOGY.......................................................................................12
3.0 Introduction.........................................................................................................................12
3.1 Research Philosophy and Design.........................................................................................12
3.2 Literature Review Methodology..........................................................................................13
3.3 Dataset Selection.................................................................................................................14
3.4 Data Preprocessing..............................................................................................................14
3.5 Model Architecture..............................................................................................................15
3.6 Model Optimisation.............................................................................................................16
3.7 Evaluation Metrics...............................................................................................................16
3.8 Ethical Considerations and Validity.....................................................................................17
3.9 Summary..............................................................................................................................17
Reference.......................................................................................................................................18
CHAPTER 1: INTRODUCTION
, 3
1.1 Background
The Internet of Things (IoT) is the fast-evolving network of physical devices that collect
and analyze data and communicate via the internet. These devices are used in various
applications such as healthcare, industry, home, and city. For instance, IoT devices are used to
track patient health, monitor household energy consumption, and gather data from manufacturing
systems to control rooms. Therefore, IoT is a critical component of the digital ecosystems today,
providing automation, efficiency, and real-time decision-making capabilities in different sectors.
IoT is becoming more prevalent because of a range of factors, including the cost of hardware
decreasing, wireless communication improving, and the need for automation. It has been
predicted that over 15 billion IoT devices are now connected, and this will increase exponentially
in the coming years (Cisco, 2023). But even though this is increasing, most IoT devices are
resource-limited. IoT devices are resource-constrained compared to traditional computing
devices and have limited memory, processing power, and energy.
These pose cybersecurity challenges. To counteract these risks, existing security
solutions, such as Intrusion Detection Systems (IDS), are built to run on powerful computing
devices and cannot run on IoT devices. Meanwhile, the threat environment has shifted. A recent
study analyzed a significant increase in IoT attacks, with over 400% growth from 2022 to 2024
(ENISA, 2024). Hackers can now use AI and automation tools to hack devices in minutes. In
addition to technical challenges, new regulations have also increased the urgency of IoT security.
The EU Cyber Resilience Act (2024) requires security by design, including detection on the
device. This shift is because it is expected that IoT devices can detect threats themselves rather
than just in the cloud. This requires the development of lightweight, efficient, and deployable
Intrusion Detection Systems (IDS) for IoT.
MSc Dissertation
Lightweight Deep Learning–Based Intrusion Detection
for Resource-Constrained IoT Devices
Student Name: Arya Vishnu
Course: MSc Cyber Security and Forensic Information Technology
Academic Year: 2025–2026
Supervisor: Peter Bednar
Chapters 1–3 Draft
, 2
Table of Contents
CHAPTER 1: INTRODUCTION............................................................................................................2
1.1 Background............................................................................................................................3
1.2 Problem Statement................................................................................................................4
1.3 Research Aim and Objectives................................................................................................4
1.4 Research Question and Hypotheses......................................................................................5
1.5 Significance, Scope, and Limitations......................................................................................5
CHAPTER 2: LITERATURE REVIEW....................................................................................................6
2.0 Introduction...........................................................................................................................7
2.1 IoT Security Challenges..........................................................................................................7
2.2 Intrusion Detection System Types.........................................................................................8
2.3 Deep Learning in Intrusion Detection....................................................................................9
2.4 Lightweight Deep Learning Techniques...............................................................................10
2.5 Research Gap.......................................................................................................................10
CHAPTER 3: RESEARCH METHODOLOGY.......................................................................................12
3.0 Introduction.........................................................................................................................12
3.1 Research Philosophy and Design.........................................................................................12
3.2 Literature Review Methodology..........................................................................................13
3.3 Dataset Selection.................................................................................................................14
3.4 Data Preprocessing..............................................................................................................14
3.5 Model Architecture..............................................................................................................15
3.6 Model Optimisation.............................................................................................................16
3.7 Evaluation Metrics...............................................................................................................16
3.8 Ethical Considerations and Validity.....................................................................................17
3.9 Summary..............................................................................................................................17
Reference.......................................................................................................................................18
CHAPTER 1: INTRODUCTION
, 3
1.1 Background
The Internet of Things (IoT) is the fast-evolving network of physical devices that collect
and analyze data and communicate via the internet. These devices are used in various
applications such as healthcare, industry, home, and city. For instance, IoT devices are used to
track patient health, monitor household energy consumption, and gather data from manufacturing
systems to control rooms. Therefore, IoT is a critical component of the digital ecosystems today,
providing automation, efficiency, and real-time decision-making capabilities in different sectors.
IoT is becoming more prevalent because of a range of factors, including the cost of hardware
decreasing, wireless communication improving, and the need for automation. It has been
predicted that over 15 billion IoT devices are now connected, and this will increase exponentially
in the coming years (Cisco, 2023). But even though this is increasing, most IoT devices are
resource-limited. IoT devices are resource-constrained compared to traditional computing
devices and have limited memory, processing power, and energy.
These pose cybersecurity challenges. To counteract these risks, existing security
solutions, such as Intrusion Detection Systems (IDS), are built to run on powerful computing
devices and cannot run on IoT devices. Meanwhile, the threat environment has shifted. A recent
study analyzed a significant increase in IoT attacks, with over 400% growth from 2022 to 2024
(ENISA, 2024). Hackers can now use AI and automation tools to hack devices in minutes. In
addition to technical challenges, new regulations have also increased the urgency of IoT security.
The EU Cyber Resilience Act (2024) requires security by design, including detection on the
device. This shift is because it is expected that IoT devices can detect threats themselves rather
than just in the cloud. This requires the development of lightweight, efficient, and deployable
Intrusion Detection Systems (IDS) for IoT.
