Secure Interval Estimation for Event-Triggered Cyber-Physical Systems Under Stealthy Attacks

Abstract

This paper presents a secure interval estimation method for nonlinear cyber-physical systems (CPSs) suffering from adversary stealthy attacks. A dynamic event-triggered scheme (DETS) is deployed in the estimation framework to reduce the unnecessary waste of limited communication resources. Considering the cases of system information fully/partially known by adversary, a state estimator is constructed based on the event-triggered system output contained by adversary. A linear parameter-varying-based (LPV-based) technique is adopted to transform the augmented error dynamic with nonlinearities into an LPV form. With considering the property of stealthy attacks, the amplitude-boundedness of attack signal is deduced, based on which a coordinated design criterion of DETS and observer is presented such that the augmented error dynamic is input-to-state stable with a prescribed $\ell _{\infty }$ performance. Then, a zonotope-based secure interval estimation algorithm is designed with considering the uncertainties induced by process disturbance, measurement noise, nonlinearities, DETS protocol, as well as malicious attacks. Finally, simulation results on a vehicle lateral dynamic system show the effectiveness and resilience of the proposed interval estimation method.