Lightweight privacy-preserving federated deep intrusion detection for industrial cyber-physical system

The emergence of Industry 4.0 entails extensive reliance on industrial cyber-physical systems (ICPS). ICPS promises to revolutionize industries by fusing physical systems with computational functionality. However, this potential increase in ICPS makes them prone to cyber threats, necessitating effective intrusion detection systems (IDS) systems. Privacy provision, system complexity, and system scalability are major challenges in IDS research. We present FedSecureIDS, a novel lightweight federated deep intrusion detection system that combines CNNs, LSTMs, MLPs, and federated learning (FL) to overcome these challenges. FedSecureIDS solves major security issues, namely eavesdropping and man-in-the-middle attacks, by employing a simple protocol for symmetric session key exchange and mutual authentication. Our Experimental results demonstrate that the proposed method is effective with an accuracy of 98.68%, precision of 98.78%, recall of 98.64%, and an F1-score of 99.05% with different edge devices. The model is similarly performed in conventional centralized IDS models. We also carry out formal security evaluations to confirm the resistance of the proposed framework to known attacks and provisioning of high data privacy and security.