Multiresolutional Quasi-Monte Carlo-based particle filters

Quasi-Monte Carlo (QMC)-based particle filters can obtain more accurate estimation than the general particle filters, with formidable computational complexity, however. Spatial-domain multiresolutional particle filters are more efficient by reducing the number of particles, but unevenly samples may cause estimation error. Aiming at these, we combine QMC numerical technique and multiresolutional methodology to improve the accuracy of filtering and computational efficiency. According to the idea, two QMC-based particle filters using thresholded wavelets in the spatial domain are proposed in this paper. The simulation shows that both the algorithms reduce the number of particles, meanwhile maintaining the estimation performance of particle filters with QMC methodology.