Asynchronous Resilient Filtering for Singular Perturbation Stochastic Semi-Markov CPSs Against Stochastic Cyber Attacks

Abstract

The article investigates the problem of asynchronous resilient filter design for cyber-physical systems (CPSs) under multiple cyber-attacks, which are characterized by singular perturbation and stochastic semi-Markov jump systems (SPSS-MJSs) with a dynamic-memory event-triggered (DMET) scheme. Firstly, we consider the vulnerability of communication network and establish a unified cyber-attack framework to address the occurrences of deception attacks and denial-of-service (DoS) attacks, simultaneously. Then, we introduce the function of the memorizer and propose the DMET scheme, which can effectively save network resources in bandwidth-limited environments. Since the mode information of the resilient filter is often unavailable, we employ a hidden Markov model (HMM) to describe the asynchronous relationship between the modes of the original system and the constructed filter. Furthermore, we judiciously select a stochastic Lyapunov functional concerning the internal dynamic variable, which results in the stochastically admissible criterion for filtering error systems under complex cyber-attacks. The co-design approach is adopted to obtain the gains of the asynchronous resilient filter and the DMET scheme. Finally, a continuous stirred-tank reactor (CSTR) system is offered to validate the effectiveness of the developed method.