Probability-Constrained Multisensor Distributed Fusion Filtering for Cyber-Physical Systems

Abstract

This article makes one of the first few attempts to investigate the multisensor distributed fusion filtering problem for a special type of time-varying nonlinear stochastic cyber-physical systems (CPSs) via encoding–decoding strategy (EDS) within the finite-horizon probability constraint framework. The random EDS is employed to orchestrate the data transmissions between sensors and remote local filters to enhance the resource-utilization efficiency and data security. A novel probability-constrained distributed fusion filtering (DFF) scheme is established such that the prescribed probabilistic ellipsoidal constraints and stochastic $H_{\infty }$ disturbance attenuation index are satisfied for the resultant local and fusion filtering errors. Sufficient conditions are firstly presented to guarantee the existence of desired local filters by iteratively solving a sequence of matrix inequalities. Subsequently, the derived multisensor distributed fusion filter is designed by means of a certain optimization problem to maximize the ellipsoidal set constraint probability of the fused filtering error. Finally, a numerical example demonstrates the validity of the proposed distributed fusion filtering approach.