Dual-Mode History-Dependent Attack Strategy Design for Cyber-Physical Systems

This paper addresses the problem of designing a dual-mode history-dependent attack strategy for cyber-physical systems (CPSs) modeled by Markov jump systems (MJSs). A crucial challenge lies in the lack of dynamic adaptability in conventional attack strategies, which rely solely on single-time instant mode information, making it difficult to strike a balance between attack effectiveness and stealthiness. To tackle this issue, virtual systems are first established to quantify the attack impacts on asynchronous observers and attack detector, integrating multi-time instant mode evolution information of both the system and observer. Unlike the existing schemes, a dual-mode history-dependent attack strategy is proposed, which exploits historical statistical characteristics of modes to enhance robustness against complex jump behaviors. Furthermore, an$H\_{-}/H\_{\infty }$optimization framework is introduced to maximize the attack's disruptive effect on the observer's estimation performance while minimizing the risk of detection. Finally, theoretical analysis and simulation results demonstrate that the proposed attack strategy achieves superior attack effectiveness and stealthiness compared to the traditional static attack models, thereby validating its practical applicability and advantages.