Distributed state estimation for heterogeneous sensor networks

This paper addresses distributed state estimation in a peer-to-peer heterogeneous sensor network characterized by varying qualities of local estimators. The proposed approach employs weighted Kullback-Leibler average of local posteriors, considering both average consensus and distributed flooding protocols to efficiently disseminate information throughout the network. Our consensus and flooding methods extend communication and fusion to include designed local weighting factors. In addition, we present a unified framework for flooding, tailored to accommodate networks with arbitrarily limited communication bandwidth. By applying these methods to average local posteriors, we derive consensus-based and flooding-based distributed state estimators. Stability of the proposed estimators is analyzed for linear systems under network connectivity and system observability. Finally, simulation results demonstrate the effectiveness of the proposed approach. (c) 2024 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.