Sparse attack reconstruction for cyber–physical systems via descriptor reduced-order observer

Abstract

Cyber attacks seriously threaten the security of Cyber–Physical Systems (CPSs) which are composed of cyber and physical realms. From the perspectives of control and system engineering, attacks will causally sabotage the performances of controlled (physical) plants with substantial damage. This paper focuses on the issue of secure estimation for CPSs under malicious false data injection (FDI) attacks implemented on sensor and actuator components synchronously. Initially, by establishing an adaptive counteraction searching strategy, the combinational entry mode of FDI attacks can be well located. Followed by which, a descriptor reduced-order switched observer from the lens of decoupling technique is crafted for reconstructing system states and sparse FDI attacks. The resulting secure estimation scheme can potentially avoid the sliding surface switching downside compared to the existing ones. Then, the sufficient conditions on ensuring the convergence of estimated errors and determining the adaptive switching counteraction logic are meticulously derived. Eventually, both the applicability and the superiority of our theoretical derivation are verified by the case study.