Attack-State Estimation for Cyber-Physical Systems: A Graph Theory Perspective

Cyber-physical systems in nowadays are more vulnerable to be attacked by malicious attackers, which are generally out of touched by the users or organizers of the underlying system, due to the widespread usage on the decentralized algorithms. In current paper, we investigate the attacked state recovery problem by virtue of the graph theory perspective. We especially show that a state could be corrected even though the support of the attacked vector is larger than half of the corrupted sensor nodes, which is crucial for the sparse signal recovery problem that is extensively investigated in area of compressed sensing in the literature. Additively, two numerical examples are presented to support the theoretical results.