Distributed fusion filtering for multi-sensor systems with correlated random transition and measurement matrices

Abstract

This paper is concerned with the distributed fusion estimation problem for discrete-time linear stochastic systems with measurements coming from different sensors and correlated random parameter matrices in both the state and measurement equations. At each sampling time, the random state transition parameter matrices are assumed to be correlated at the same sampling time with the measurement random parameter matrices of each sensor. Moreover, the random parameter matrices in the observation equations are one-step auto-correlated and cross-correlated between the different sensors. The additive noises are also assumed to be correlated. Under these assumptions, the distributed fusion filter is designed as the matrix-weighted linear combination of the local least-squares linear filters obtained at every single sensor, using the linear minimum variance optimality criterion. A numerical simulation example considering a two-sensor system with randomly delayed measurements is used to illustrate the applicability of multi-sensor systems with correlated random parameter matrices and analyse the performance of the proposed filtering estimators.