Anomaly Detection and Identification for Multiagent Systems Subjected to Physical Faults and Cyberattacks

This article investigates the anomaly detection and identification problem for multiagent systems subjected to both physical faults in facilities and false-data-injection attacks in communication networks. A novel secure scheme composed of independent and cooperative detectors is proposed to detect and identify anomalies. To improve the reliability of the detection results, the $H_\infty$ multiobjective optimization approach is applied where a compromise between sensitivity to anomalies and robustness to disturbances is generated. Moreover, the cooperation mechanism of the cooperative detector is developed based on edges' betweenness centrality to further improve the agents' detection performance. Effectiveness and improvements of the proposed scheme are validated on a multivehicle experimental platform.