Distributed fusion estimation with square-root array implementation for Markovian jump linear systems with random parameter matrices and cross-correlated noises

This study presents the distributed fusion estimation of discrete-time Markovian jump linear systems with random parameter matrices and cross-correlated noises in sensor networks. The recursive linear minimum mean square error estimator is proposed based on the Gram-Schmidt orthogonalization procedure under a centralized framework. In order to avoid the loss of positive semidefiniteness and reduce dynamical range, its square-root array implementation is presented by recursively triangularizing the square roots of relevant positive semidefinite matrices. Furthermore, via the information filter form, the distributed fusion estimation with square-root array implementation is derived from the centralized fusion structure, incorporated with consensus strategy. A maneuvering target tracking simulation in a sensor network validates the proposed method.