Globally Optimal Distributed State Fusion Filter for Systems with Random Parameter Matrices and Correlated Noises

This paper is concerned with the globally optimal state fusion estimation problem for multi-sensor systems with random parameter matrices and correlated process and measurement noises. Firstly, the considered random parameter system is transformed into a stochastic uncertain system with multiple multiplicative noises by introducing the matrix unit. Next, the multiplicative noises are converted into the virtual additive noises. Then, based on the projection theory and difference technology, a globally optimal distributed state fusion (DSF) filter without feedback in the sense of linear minimum variance (LMV) is proposed. The proposed DSF filter has the same estimation accuracy as the centralized fusion (CF) one. A tracking system example verifies the effectiveness of the proposed algorithm.