Resilient Distributed Filtering for Discrete Time-varying Systems with Missing Measurements and Stochastic Uncertainties

Abstract

In this paper, the resilient distributed filtering problem is addressed for discrete stochastic uncertain time-varying systems with missing measurements and stochastic uncertainties over wireless sensor networks. Some random variables governed by the Bernoulli distribution are used to model the missing measurements phenomenon of each sensor node. In addition, the stochastic uncertainties are characterized by the multiplicative noises and stochastic nonlinearities. By using the variance-constrained method, an appropriate filter gain is selected to minimize the trace of the upper bound for the filter error covariance. Moreover, the resilient distributed filtering algorithm is designed and a new matrix simplification technique is introduced to deal with the sparsity of sensor networks topology. Finally, both the feasibility and effectiveness of the resilient distributed filtering algorithm are verified by a numerical simulation.