Improved Weighting in Particle Filters Applied to Precise State Estimation
This letter presents a methodology to reduce the computational burden of Particle Filters (PF) used to achieve a target state estimation accuracy in Bayesian tracking problems. A strategy, named Multiple Weighting (MW), that exploits information diversity of the input measurements for an efficient weighting of the particles is introduced. This study shows that, by relying on the prior knowledge of the state-observations relationships, it is possible to achieve a significant reduction in the number of samples required to obtain any target accuracy. An experimental assessment is provided with an application to precise positioning estimation in Global Navigation Satellite System (GNSS) receivers.