Markov Chain Monte Carlo Particle Algorithms for Discrete-Time Nonlinear Filtering Academic Article uri icon

abstract

  • This work shows how a carefully designed instrumental distribution can im-prove the performance of a Markov chain Monte Carlo (MCMC) particle filter for systems with a high state dimension (up to 100). We devise a special subgradient-based kernel from which candidate moves are drawn. This facilitates the implementation of the filter-ing algorithm in high dimensional settings using a remarkably small number of particles. We demonstrate our approach in solving a nonlinear non-Gaussian dynamic compressed sensing (l 1 constrained) problem and show high estimation accuracy.

publication date

  • January 1, 2013