Distributed Filtering for Switched Linear Systems With Sensor Networks in Presence of Packet Dropouts and Quantization

Abstract

This paper is concerned with the distributed $H_\infty $ filtering problem of discrete-time switched linear systems in sensor networks in face of packet dropouts and quantization. Specifically, due to the packet dropout phenomenon, the filters may lose access to the real-time switching signal of the plant. It is assumed that the maximal packet dropout number of switching signal is bounded. Then, a distributed filtering system is proposed by further considering the quantization effect. Based on the Lyapunov stability theory, a sufficient condition is obtained for the convergence of filtering error dynamics. The filter gain design is transformed into a convex optimization problem. In this paper, a quantitative relation between the switching rule missing rate and filtering performance is established. Furthermore, the upper bound of the switching rule missing rate is also calculated. Finally, the effectiveness of the proposed filter design is validated by a simulation study on the pulse-width-modulation-driven boost converter circuit. The impact of noise covariance, system dynamics, and network connectivity is studied, and some discussions are presented on how these parameters affect the filtering performance.