Security Metrics and Tradeoff of Cyber-Physical Systems Subject to False Data Injection Attacks

This paper focuses on the vulnerable and stealthy regions of the intricate cyber attackers pertaining to state-of-the-art cyber-physical systems from the perspective of control-theoretic metrics. Using the minimum-energy attack signals and the L₂-gain of output signals as vulnerability and stealthiness, respectively, these two predominant concerns of attackers and the corresponding vulnerable subspaces and stealthy subspaces are interpreted by the eigenvalues and eigenvectors of controllability Gramian matrix and observability Gramian matrix. It is also shown that the principal angles of these two subspaces indicate the tradeoff between the vulnerability and stealthiness of subspaces. A model of moving vehicles validates the effectiveness of theoretical results. © The Institution of Engineering & Technology 2023.