ºìÐÓÊÓƵ

Topic

AI-Driven Security in Software-Defined Networks: A Unified Framework for Intrusion Detection and Mitigation

Department of Computer Science

Date & location

• Tuesday, April 29, 2025

• 11:00 A.M.

• Virtual Defence

Reviewers

Supervisory Committee

• Dr. Sudhakar Ganti, Department of Computer Science, University of Victoria (Supervisor)

• Dr. Hausi Muller, Department of Computer Science, UVic (Member)

• Dr. Issa Traore, Department of Electrical and Computer Engineering, UVic (Outside Member) 

External Examiner

• Dr. Nur Zincir-Heywood, Faculty of Computer Science, Dalhousie University 

Chair of Oral Examination

• Dr. Pauline van den Driessche, Department of Mathematics and Statistics, UVic 

Abstract

Over the past decade, data networks have evolved from static resource deployment to a more dynamic and adaptive paradigm. Software-Defined Networking (SDN) is one of the most creative network technologies where network control is separated from forwarding. It is directly programmable and has been proposed as a way to programmatically control networks, facilitating the deployment of new applications and services, as well as tuning network policies and performance. However, various challenges have hindered achieving strong cybersecurity within the dynamic network configurations of Software-Defined Networking. Traditional cybersecurity measures, especially in programmable and dynamic network infrastructures like SDNs, are not sufficient to mitigate cyber threats. This dissertation explores the capabilities of SDN and examines how AI-driven methods can enhance intrusion detection and mitigation. The study begins by providing a comprehensive introduction to SDN, outlining its fundamental capabilities and comparative advantages over traditional network architectures. In addition, it explores SDN vulnerabilities and addresses complex security challenges.

The objective of this thesis work is to improve the detection and mitigation of threats in SDN environments. For this, we first present a dynamic defense framework that includes Machine Learning and Deep Learning techniques for attack detection and mitigation. Furthermore, a novel hybrid Coot-Lyrebird optimization algorithm is developed to specifically choose the most impactful features in the network. The selected features are given to the proposed hybrid network that combines Convolutional Neural Network (CNN), SE-ResNeXt, and Long Short-Term Memory (LSTM) networks. Finally, the proposed Deep Q-Network (DQN) model performs attack mitigation measures. The results indicate that the proposed dynamic defense has an accuracy of 0.999571%. In addition, we extended our study to include more complex environments. Software-Defined Internet of Things (SD-IoT) networks enabled intelligent network management through their dynamic features, but expose centralized infrastructure to complex cyberattacks that put the system in great danger. In order to address this, a novel federated secure intelligent intrusion detection and mitigation framework with automated attack reporting for SD-IoT network is presented.