Delay-Tolerant Data Fusion for Underwater Acoustic Tracking Networks

Abstract

We consider a network of distributed underwater sensors whose task is to monitor the movement of objects across an area. The sensors measure the strength of signals emanated by the objects and convey the measurements to the local fusion centers. Multiple fusion centers are deployed to cover an arbitrarily large area. The fusion centers communicate with each other to achieve consensus on the estimated locations of the moving objects. We introduce two efficient methods for data fusion of distributed partial estimates when delay in communication is not negligible. We concentrate on the minimum mean squared error (MMSE) global estimator, and evaluate the performance of these fusion methods in the context of multiple-object tracking via extended Kalman filtering. Numerical results show the superior performance compared to the case when delay is ignored.