Finite frequency distributed fault detection in sensor networks with memory event-triggered scheme and deception attacks

Abstract

This paper concentrates on the finite frequency distributed fault detection (FD) problem for discrete-time nonlinear switched systems (SSs) with time-varying delays through sensor networks. First, the switching mechanism is described by the so-called sojourn probability method. Next, a distributed memory event-triggered scheme (METS) is employed to alleviate the network bandwidth load, and the adverse effects of malicious deception attacks are considered and analyzed. Furthermore, a finite frequency distributed FD filter (FDF) subject to METS and deception attacks is developed corresponding to each sensor node, and the existence conditions of the FDF are offered in the form of linear matrix inequalities. Lyapunov stability analysis shows that the resulting augmented filter system is mean-square stable and meets the specified full-frequency $H_{\infty }$ level bound and finite-frequency $H_{-}$ level bound. Comparative simulations demonstrate that the proposed finite-frequency distributed FDF can detect unknown faults in nonlinear SSs on sensor networks more quickly and accurately than some existing full-frequency solutions.