Resilient Observer-Triggered Adaptive Control for Cyber-Physical Systems Under Time-Vary Stealthy FDI Attacks

Abstract

Cyber-physical systems (CPSs) are widely used in safety-critical applications, where both control reliability and communication efficiency are essential. However, open networks make CPSs vulnerable to false data injection (FDI) attacks, which threaten system stability. Existing event-triggered control methods often fail to simultaneously ensure attack resilience, stability, and $H_\infty$ performance. This paper addresses the secure control problem of CPSs under FDI attacks by proposing an observer-based dynamic event-triggered control framework. To counteract the adversarial disturbances, a novel attack-resilient observer is designed to simultaneously estimate both the system states and the injected attack signals, enabling the synthesis of a secure observer-based controller. An advanced dynamic event-triggered mechanism (DETM) is developed by incorporating an internal dynamic variable, which adaptively adjusts triggering thresholds to significantly reduce communication frequency while avoiding Zeno behavior. Through Lyapunov-Razumikhin analysis, the closed-loop system is proven to achieve asymptotic stability and guaranteed $H_\infty$ performance, ensuring robustness against bounded FDI attacks. Theoretical results are validated via numerical simulations, demonstrating the effectiveness of the proposed method in mitigating attack impacts and conserving network resources.