Complexity Analysis of the Marginalized Particle Filter
2004 (English)Report (Other academic)
In this paper the computational complexity of the marginalized particle lter is analyzed. We introduce an equivalent flop measure to capture floating-point operations as well as other features, which cannot be measured using flops, such as the complexity in generating random numbers and performing the resampling. From the analysis we conclude how to partition the estimation problem in an optimal way for some common target tracking models. Some guidelines on how to increase performance based on the analysis is also given. In an extensive Monte Carlo simulation we study different computational aspects and compare with theoretical results.
Place, publisher, year, edition, pages
Linköping: Linköping University Electronic Press, 2004. , 15 p.
LiTH-ISY-R, ISSN 1400-3902 ; 2611
Nonlinear estimation, Marginalized particle filter, Rao-Blackwellization, Kalman filter, Complexity analysis, Equivalent
IdentifiersURN: urn:nbn:se:liu:diva-55989ISRN: LiTH-ISY-R-2611OAI: oai:DiVA.org:liu-55989DiVA: diva2:316745