The optimality for the distributed Kalman filtering fusion with feedback (Q5947660)

This contribution deals with Kalman filtering for multi-sensor systems where each local sensor carries out its own state estimation before it is submitted to a fusion center. The fusion center calculates the state estimation based on the field information. Such a procedure is used to increase the system survivability. The authors expose the advantages of the feedback where the fusion center broadcasts the recent full-information estimation to all sensors to update them. It is shown that the distributed filtering with feedback achieves the same performance as centralized filtering. The main result is that feedback increases the local tracking performance of the distributed Kalman filtering.