A variational Bayesian multiple particle filtering scheme for large-dimensional systems
KAUST DepartmentApplied Mathematics and Computational Science Program
MetadataShow full item record
AbstractThis paper considers the Bayesian filtering problem in high-dimensional nonlinear state-space systems. In such systems, classical particle filters (PFs) are impractical due to the prohibitive number of required particles to obtain reasonable performances. One approach that has been introduced to overcome this problem is the concept of multiple PFs (MPFs), where the state-space is split into low-dimensional subspaces and then a separate PF is applied to each subspace. Remarkable performances of MPF-like filters motivated our investigation here into a new strategy that combines the variational Bayesian approach to split the state-space with random sampling techniques, to derive a new computationally efficient MPF. The propagation of each particle in the prediction step of the resulting filter requires generating only a single particle in contrast with standard MPFs, for which a set of (children) particles is required. We present simulation results to evaluate the behavior of the proposed filter and compare its performances against standard PF and a MPF.
CitationA variational Bayesian multiple particle filtering scheme for large-dimensional systems 2016:1 IEEE Transactions on Signal Processing
SponsorsResearch reported in this publication was supported by King Abdullah University of Science and Technology (KAUST). We would like to thank three anonymous reviewers for their constructive comments and suggestions.