Parallel Consensus on Likelihoods and Priors for Networked Nonlinear Filtering

Abstract

A novel consensus approach to networked nonlinear filtering is introduced. The proposed approach is based on the idea of carrying out in parallel a consensus on likelihoods and a consensus on prior probability distributions and then combine the outcomes with a suitable weighting factor. Simulation experiments concerning a target tracking case-study show that the proposed consensus-based nonlinear filter can be convenient when only a few consensus iterations per sampling interval can be afforded.