Interacting multiple model particle filtering using new particle resampling algorithm

Dah Chung Chang, Meng Wei Fan

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations

Abstract

The state estimation technique based on the Kaiman filter (KF) is widely used in many communication applications. The KF is only optimal for linear modeling with independent and identically distributed (i.i.d.) random variables and Gaussian noises. In some complicated problems, the system model is not unique and the measurement equation is nonlinear. The particle filter (PF) along with interacting multiple models (IMM) becomes an attractive solution. In this paper, a new particle resampling method is proposed for the PF to alleviate the degeneracy effect of particle propagation. The new IMMPF algorithm is developed for an angle-of-arrival (AOA) tracking problem with bearings-only measurements. Simulation results show that the IMMPF algorithm outperforms the IMM extended KF algorithm and achieves a root mean square tracking performance which is quite close to the posterior Cramer-Rao lower bound (CRLB).

Original languageEnglish
Title of host publication2014 IEEE Global Communications Conference, GLOBECOM 2014
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3215-3219
Number of pages5
ISBN (Electronic)9781479935116
DOIs
StatePublished - 9 Feb 2014
Event2014 IEEE Global Communications Conference, GLOBECOM 2014 - Austin, United States
Duration: 8 Dec 201412 Dec 2014

Publication series

Name2014 IEEE Global Communications Conference, GLOBECOM 2014

Conference

Conference2014 IEEE Global Communications Conference, GLOBECOM 2014
Country/TerritoryUnited States
CityAustin
Period8/12/1412/12/14

Keywords

  • IMM
  • Kaiman filtering
  • particle filtering
  • posterior CRLB
  • resampling
  • State estimation

Fingerprint

Dive into the research topics of 'Interacting multiple model particle filtering using new particle resampling algorithm'. Together they form a unique fingerprint.

Cite this