Federated learning-based intrusion detection system for industrial Internet of Things: enhancing security and efficiency

As the Industrial Internet of Things (IIoT) rapidly evolves, cybersecurity issues have become increasingly prominent. Traditional centralized intrusion detection methods face significant challenges, including privacy, security, and computational resource limitations with large-scale heterogeneous data. This paper proposes a federated learning-based intrusion detection method, combining Convolutional Neural Networks (CNN) and Gated Recurrent Units (GRU) with Isolation Trees for anomaly detection and removal. This approach addresses the non-independent and identically distributed (non-IID) data in IIoT and provides personalized local model training. Experimental results show that the proposed method significantly improves intrusion detection accuracy and real-time performance.