GraphFedAI framework for DDoS attack detection in IoT systems using federated learning and graph based artificial...

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• The Internet of Things (IoT) presents significant security and privacy challenges, particularly distributed denial-of-service (DDoS) attacks.
• Traditional DDoS detection methods struggle with privacy, scalability, and adaptability in dynamic IoT environments.

Purpose of the Study:

• To propose GraphFedAI, a novel framework for robust, scalable, and privacy-preserving DDoS detection in heterogeneous IoT networks.
• To address the limitations of conventional methods by integrating advanced techniques for enhanced security.

Main Methods:

• Utilizing adaptive session-based graph modeling to represent IoT networks as dynamic graphs.
• Employing Pearson correlation-guided feature selection and interpolation-aware graph neural network (GNN) training.
• Incorporating federated learning (FL) for privacy-preserving, localized model training and enhanced scalability.

Main Results:

• GraphFedAI demonstrated high detection accuracy for DDoS attacks in dynamic IoT conditions.
• The framework achieved low false positive rates, indicating reliable performance.
• Evaluations confirmed strong resilience and effectiveness on the CIC-IoT-2023 dataset.

Conclusions:

• GraphFedAI offers a significant advancement in securing IoT networks against DDoS attacks.
• The framework's integration of GNNs and federated learning provides a scalable and privacy-preserving solution.
• This approach enhances the overall security posture of heterogeneous IoT ecosystems.