Non-linear and non-Gaussian state estimation using log-homotopy based particle flow filters

Non-linear filtering is a challenging task and generally no analytical solution is available. Sub-optimal methods like particle filters are employed to approximate the conditional probability densities. These methods are expensive in terms of the processing requirements. Recently proposed log homotopy based particle flow filter, also known as Daum-Huang filter (DHF) provides an alternative way of non-linear state estimation. Based on different assumptions, several versions of DHF have been derived. Superior performance has been reported for their use in several non-linear but Gaussian filtering problems. In this paper we compare the performance of different versions of DHF for a coupled, non-linear and non-Gaussian system model. Results show that recently proposed non zero diffusion DHF perform better than previous versions of DHF.