A New Nonlinear Filter Algorithm Based on QMC Quadrature

In order to avoid the possible gaps and clusters that arise from random sampling in Monte Carlo (MC) methods, and improve the sampling efficiency and calculation accuracy, the Quasi-Monte Carlo (QMC) methods are to be applied to replace it. The idea in QMC is to use more regularly distributed and deterministic points for sampling an integrand. We propose a new nonlinear filter by applying the QMC sampling methods to the particle filter algorithm. Given certain proposal distributions, a simulation example is presented. The results show that the nonlinear filter based on the QMC methods performs more efficient than that based on the MC methods. The performance provides some references for the real-time application of particle filter in nonlinear / non-Gaussian systems.