A Covert Cyber‐Attack Framework and Detection on Communication Links in Linear Multi‐Agent Cyber‐Physical Systems

ABSTRACT This study presents covert attack architecture and detection approach in communication networks for homogenous and linear multi‐agent systems (MASs). This work focuses on distributed MASs with bounded process and measurement disturbances. A residual‐based detection method, exploiting a KALMAN filter observer and extra dynamic, is proposed to detect covert attacks on agents' communication channels. On the basis of the methodology, each agent or subsystem has its own anomaly detector which allows detecting the presence of covert attack using only local information. Finally, simulation results on the consensus problem of MAS are investigated. The results illustrate the effectiveness of the proposed approach for the detection of covert attack in a distributed manner.