Stable Remote Optimal Filtering and Fusion With Communication Driven by Cumulative Estimate Innovation

This article studies remote filtering for a discrete-time linear system observed by one or more sensors with limited communication resources. We propose new communication schemes based on cumulative estimate innovation and derive the corresponding minimum mean square error (MMSE) filter and fuser. Our communication schemes balance communication cost and estimation performance. The proposed remote filter and fuser can improve estimation under limited communication resources by leveraging the information of no transmission. Further, it is proved that these filters have guaranteed stability - the expected norms of the mean square error (MSE) matrices are bounded and their upper bounds are obtained. We also give the transmission rate, conditional probabilities of a future transmission and conditional probabilities based on the elapsed time since the latest transmission. The effectiveness of the proposed methods is illustrated by simulation examples.