Distributed filtering based on weighted average strategy in unreliable sensor networks

This paper studies the distributed filtering problem of heterogeneous sensor networks. The communications among sensors and the sensing links are unreliable and randomly lost. Based on Kalman filtering algorithm and weighted average strategy, a sub-optimal filtering algorithm is proposed. The statistical convergence properties of estimation error covariances are investigated and a necessary and sufficient convergence condition is proposed based on LMIs. A special network with sensing link losses is further studied and an explicit necessary convergence condition concerning the observability of sensors and sensing link loss probability is given. Simulation examples are given to illustrate the results.