Approximate conditional mean particle filter

Author

Yee, Derek ; Reilly, James P. ; Kirubarajan, Thia

Author_Institution

Dept. of Electr. & Comput. Eng., McMaster Univ., Hamilton, Ont.

fYear

2005

fDate

17-20 July 2005

Firstpage

405

Lastpage

410

Abstract

We consider partially observed non-Gaussian dynamic state space models in which the process equation consists of a combination of linear and nonlinear states and the process noise for the nonlinear state update is distributed according to a mixture of Gaussians. In this paper, we solve a Bayesian filtering problem. The proposed filter is an efficient combination of the particle filter and the approximate conditional mean filter. Simulation results on a time-varying autoregressive signal demonstrate the effectiveness of the proposed algorithm

Keywords

Bayes methods; Gaussian distribution; approximation theory; autoregressive processes; nonlinear filters; particle filtering (numerical methods); time-varying filters; Bayesian filtering problem; Gaussian mixture; conditional mean particle filter approximation; dynamic state space model; time-varying autoregressive signal; Bayesian methods; Decision support systems; Filtering; Gaussian distribution; Gaussian noise; Nonlinear equations; Particle filters; Signal processing algorithms; State-space methods; White noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Statistical Signal Processing, 2005 IEEE/SP 13th Workshop on

Conference_Location

Novosibirsk

Print_ISBN

0-7803-9403-8

Type

conf

DOI

10.1109/SSP.2005.1628629

Filename

1628629