Distributed filtering for T-S fuzzy systems under cyber-attacks with time-varying saturation function

Abstract

This paper focuses on the distributed filter issue for a class of nonlinear systems under hybrid cyber-attacks, encompassing both deception attacks and denial of service (DoS) attacks with uncertain attack probabilities. In the sensor network, each filter estimates the output signals of the systems by dealing with the output measurements from the systems and the information received from its neighbors and each filter may receive the state estimation information affected by outliers from its neighbors, which is propagated based on the communication topology. With the purpose of mitigating the effects from channel noise during the signal transmission on filtering error systems (FES) and the potential anomalies’ impact in signal transmission caused by the hybrid cyber-attacks, a dynamic saturation function-based distributed filter is designed during the filtering process, whose saturation level is adaptively varying based on previous measurement errors. Through this approach and by utilizing Lyapunov-Krasovskii theory, sufficient conditions are established to ensure the stochastic stability (SS) of the FES and to achieve the predefined H∞ performance objectives. Finally, a practical model is presented to demonstrate the effectiveness and practicality of the designed distributed filter methodology.