Graph-Based Anomaly Detection in Large-Scale Industrial IoT Network

Abstract

The rapid proliferation of Industrial Internet of Things (IIoT) networks has revolutionized industrial automation and monitoring. However, the increasing complexity and scale of these networks introduce significant security and operational challenges, particularly in detecting anomalous behaviors that may indicate cyber threats or system failures. Traditional anomaly detection methods often struggle with the dynamic and high-dimensional nature of IIoT data. This study explores a graph-based anomaly detection framework that leverages graph neural networks (GNNs) and spectral analysis to identify deviations in IIoT network traffic and device interactions. By modeling IIoT networks as dynamic graphs, the proposed method captures both structural and temporal dependencies, enabling more accurate and scalable anomaly detection. Experimental evaluations on real-world IIoT datasets demonstrate that the proposed framework outperforms traditional machine learning and deep learning approaches in terms of detection accuracy, false positive rate, and computational efficiency. The findings suggest that graph-based techniques offer a robust solution for securing large-scale IIoT infrastructures against evolving cyber threats and operational anomalies.