Dynamic Event-Triggered Distributed Sequential Consensus Fusion Filtering for Sensor Networks

Abstract

This article investigates the distributed consensus filtering problem in sensor networks and proposes the optimal distributed sequential consensus fusion filtering (DSCFF) algorithm. Each sensor node in the network sequentially exchanges information with its neighboring nodes over multiple rounds to obtain global information. The filtering results for all sensor nodes tend to agree, but significant information is repeatedly exchanged between individual nodes, consuming the limited energy in the network. A dynamic event-triggering (DET) mechanism based on the minimum covariance per round is proposed to reduce unnecessary energy loss and decrease the communication bandwidth between sensor nodes. In addition, as the optimal DETDSCFF needs to calculate the cross-covariance matrices (CCMs) between sensor nodes, which increases the calculation complexity, this article provides the suboptimal DETDSCFF algorithm that minimizes the upper bound of the error covariance during fusion to solve the consensus gain. The boundedness of this suboptimal filter is proven, and its effectiveness is proven through simulation experiments.