A distributed particle-PHD filter using arithmetic-average fusion of Gaussian mixture parameters

We propose a particle-based distributed PHD filter for tracking the states of an unknown, time-varying number targets. To reduce communication, local filters at neighboring sensors communicate Gaussian mixture (GM) parameters. In contrast to most existing filters, our employs “arithmetic average” fusion. For particles–GM conversion, we use method that avoids particle clustering and enables significance-based pruning GM components. GM–particles develop importance sampling based parallelization filtering dissemination/fusion operations. The resulting framework is able integrate both GM-based filters. Simulations demonstrate excellent performance small communication computation requirements filter.