Abstract

The proliferation of Internet of Things (IoT) devices has created a vast and complex attack surface, rendering traditional centralized intrusion detection systems (IDS) inadequate for effectively safeguarding these networks. This paper proposes a novel architecture for an adaptive Intrusion Detection System (IDS) that leverages federated learning (FL) and blockchain-based trust management to enhance security in IoT networks. The proposed system allows IoT devices to collaboratively train a global intrusion detection model without sharing sensitive data, preserving privacy and reducing communication overhead. A blockchain is employed to establish a decentralized trust mechanism, ensuring the integrity and reliability of the federated learning process by tracking and verifying contributions from individual devices. The system's performance is evaluated through simulations and real-world experiments, demonstrating its ability to detect a wide range of IoT-specific attacks with high accuracy and minimal false positives. The results highlight the potential of this approach to significantly improve the security posture of IoT networks while addressing key challenges related to privacy, scalability, and trust.