Consensus Tracking of an Underwater Vehicle Using Weighted Harmonic Mean Density

TL;DR

This paper introduces a weighted harmonic mean density (HMD) approach for consensus tracking of underwater vehicles, effectively fusing Gaussian densities from multiple sensors to improve tracking accuracy in a distributed sensor network.

Contribution

It proposes a novel HMD-based fusion method with optimal weighting via KL divergence minimization for underwater target tracking.

Findings

HMD-based fusion outperforms existing methods in simulations.

Optimized weights improve tracking accuracy.

The method effectively fuses Gaussian densities in distributed networks.

Abstract

This paper addresses an underwater target tracking problem in which a large number of sonobuoy sensors are deployed on a surveillance region. The region is divided into several sub-regions, where a single tracker, capable of generating track is installed. Each sonobuoy can measure the direction of arrival of acoustic signals (known as bearing angles) and communicate the measurements with the local tracker. Further, each local tracker can communicate with all other trackers, where each of them can exchange their estimate and finally a consensus is reached. We propose a weighted harmonic mean density (HMD) based tracking to reach a consensus and provide a solution for the fusion of Gaussian densities. In this approach, optimal weights are assigned by minimizing the Kullback-Leibler divergence measure. Performance of the proposed method is measured using root mean square error, percentage of track divergence, and normalized estimation error squared. Simulation results demonstrate that the optimized HMD-based fusion outperforms existing fusion methods during a distributed tracking.