Monitoring and detection of malicious adversarial zero dynamics attacks in cyber-physical systems

Abstract

This paper is mainly concerned with monitoring and detection of zero dynamics (ZD) cyber attacks that are injected by malicious hackers and adversaries to safety critical cyber-physical systems (CPS). We consider a CPS system where the physical system (i.e., the plant) is represented by a linear time-invariant dynamical process. Specifically, we propose and provide a methodology for detecting zero dynamics cyber attacks through introducing an auxiliary system and detection filters. When compared to the currently available methods in the literature, the key advantage of our proposed strategy is that even if the attacker has complete knowledge of the CPS system including knowledge of our proposed approach, the introduced auxiliary system and filters, the attacker cannot design an undetectable attack that significantly and adversely impact stability and performance of the CPS system.