Event-Triggered Distributed Hierarchical Filtering with Random Link Failures

Abstract

This paper focuses on the event-triggered distributed hierarchical filtering over wireless networks consisting of local wireless sensor networks in the first layer and wireless cluster networks in the second layer. To reduce communication consumption over local wireless sensor networks, recursive event-triggered sequential local estimators are first proposed in each cluster head by the co-design of an event-triggered scheduling mechanism and sequential local estimators, with only necessary measurements being transmitted to the cluster head in an individual cluster. A distributed hierarchical estimator is further derived by fusing local estimates over unreliable wireless cluster networks, where interacted estimates may be unavailable due to potential random link failures. It is noted that the designed hierarchical estimator has significant advantages in reducing communication bandwidth and improving execution efficiency. Finally, a target tracking example is shown to verify the proposed results.