Cauchy-Schwarz divergence-based distributed fusion with poisson random finite sets

This paper presents a new approach towards statistical fusion of multi-source information. Our solution is formulated in the context of fusing the Poisson finite random set posteriors returned by multiple local PHD filters at sensor nodes of a distributed multi-sensor multi-object estimation system. The most common measure used for information gain in stochastic multi-source information fusion is Kullback-Leibler divergence (KLD) which leads to the well-known Generalised Covariance Intersection (GCI) rule for sensor fusion. We present the idea of using Cauchy-Schwarz divergence instead of KLD and derive a closed-form solution for fusion of multiple Poisson posteriors. Simulation results show that our method performs favourably against GCI fusion rule in terms of overall tracking performance.