A new particle filter based on smooth variable structure filter

In this paper, a new particle filter (PF) based on the smooth variable structure filter (SVSF) is presented. Here, PF is a popular estimation method that makes use of distributed particles to form an approximation of the probability distribution function. The SVSF is a sliding mode estimator, which is based on predictor-corrector methods and sliding mode concepts. Moreover, the gain of the SVSF is calculated based on a switching surface. The new filter called SVSF-PF utilizes a smoothing boundary layer of SVSF to control the particles in a range, which can improve the overall estimation accuracy and stability. Finally, a simulation example of nonlinear target tracking is used to demonstrate effectiveness.