Optimal sequential Kalman filtering with cross-correlated measurement noises

The purpose of this paper is to present a new optimal sequential decentralized filtering algorithm for discrete time-varying linear control systems with cross-correlated noises. The new method uses a hierarchical structure to perform successive orthogonalization of the measurement noises, and drives the new algorithm based on the well-known projection theorem. The estimator can process the system with measurements delay as well as data missing because the update step is just according to the coming order of sensors in a recursive form. Finally, the precision relation between the new algorithm and the centralized multisensor fusion method is proved and simulation result shows that new filter is better than other similar filters in performance.