Strengthening network DDOS attack detection in heterogeneous IoT environment with federated XAI learning approach

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Network Security

Background:

• The proliferation of Internet of Things (IoT) devices increases vulnerability to Distributed Denial of Service (DDoS) attacks.
• Existing centralized security systems struggle with privacy and scalability issues in IoT environments.
• There is a critical need for robust, privacy-preserving, and scalable solutions for IoT cybersecurity.

Purpose of the Study:

• To propose and evaluate a novel method for detecting and preventing DDoS attacks in IoT networks.
• To leverage Explainable Artificial Intelligence (XAI) and Federated Deep Neural Networks (FDNNs) for enhanced threat detection.
• To address the limitations of traditional security systems in privacy and scalability for IoT.

Main Methods:

• Implementation of Federated Deep Neural Networks (FDNNs) for privacy-preserving training on distributed data.
• Integration of XGBoost with SHapley Additive exPlanations (SHAP) for feature selection and model interpretability.
• Training FDNN models over multiple rounds using diverse DDoS attack datasets from client devices.

Main Results:

• The proposed solution demonstrates high robustness, privacy preservation, and scalability for DDoS attack detection in IoT.
• Achieved a remarkable 99.78% accuracy in detecting DDoS attacks.
• Reported a precision rate of 99.80%, recall of 99.74%, and an F1 score of 99.76%.

Conclusions:

• Federated Learning (FL)-based Intrusion Detection Systems (IDS) are effective against IoT cybersecurity challenges.
• The developed XAI and FDNN approach offers a promising solution for securing modern network infrastructures.
• This method provides a scalable and privacy-conscious framework for mitigating evolving cyber threats in IoT ecosystems.