Setting sample size in particle filters using Cramer-Rao bound

Particle filter solution of state estimation for stochastic nonlinear system is addressed. Aspects of particle filter are discussed and attention is devoted namely to crucial parameter of the filter - sample size. The way of determination of sufficient sample size in particle filter is proposed. The mean square error matrices of particle filter state estimates are compared for different sample sizes and the computational demands of the filter are discussed as well. To assist in setting sample size and for filter quality evaluation, the Cramer Rao bound is used. The designed procedure is illustrated by a numerical example. Copyright (C) 2001 IFAC.