Multi-Layer Adaptive Intrusion Detection and Mitigation System for SDN Adversarial Threats using a BAT-MC Model
Sanjay Kumar Bose, G Gokulraj, N Maheswaran, G Logeswari, T Anitha, G Vijayaraj · 2024
In Software-Defined Networking (SDN), Intrusion Detection Systems (IDS) play a pivotal role in safeguarding network security by monitoring traffic for anomalies or policy violations. This paper proposes the BAT-MC model, an advanced intrusion detection and mitigation framework that integrates Bidirectional Long Short-Term Memory (BiLSTM), attention mechanisms, and convolutional dropout layers to enhance anomaly detection and traffic optimization. The model dynamically learns from adversarial attacks, leveraging iterative refinements to adapt to sophisticated threats. Using the In-SDN dataset, the BAT-MC model achieves an 85% accuracy rate in anomaly detection, demonstrating its efficacy in capturing relevant features, maintaining network reliability, and fortifying SDN environments against emerging adversarial threats. Additionally, the integration of SNORT for IP blacklisting enhances threat mitigation, ensuring real-time isolation of attackers. This research establishes the BAT-MC model as a robust and scalable solution for addressing modern cybersecurity challenges in SDN.