Author: Omair Faraj
Programme: Doctoral Programme in Network and Information Technologies
Language: English
Supervision: David Megias, Joaquin Garcia-Alfaro
Faculty / Institute: Doctoral School UOC
Subjects: Computer Science
Key words: Watermarking, Zero-watermarking, Data provenance, Internet of things, Intrusion detection, Cyber security, Machine learning
Area of knowledge: Network and Information Technologies
Abstract:
This thesis explores the integration of advanced security techniques into Intrusion Detection Systems (IDS) for IoT networks, which face increasing cyber threats due to their interconnected nature and limited resources. Traditional IDS methods, such as signature-based detection, only identify known attacks, while anomaly detection can uncover unknown attacks but often generates high false alarms. To address these challenges, we propose a robust, lightweight approach for data integrity and data provenance in IoT networks. This includes a zero-watermarking technique to secure provenance information and a two-layer IDS model that combines Machine Learning (ML) classification with zero-watermarking to enhance detection accuracy. We systematically review both ML-based IDS and data provenance security techniques, identifying challenges and open issues. Additionally, we validate our approach through security analysis, numerical simulations, and experiments, demonstrating its computational efficiency and effectiveness in enhancing IDS for IoT networks.