Gaussian proposal density using moment matching in SMC methods

S. Saha, P.K. Mandal, Y. Boers, H. Driessen, A. Bagchi

    Research output: Contribution to journalArticleAcademicpeer-review

    24 Citations (Scopus)
    22 Downloads (Pure)

    Abstract

    In this article we introduce a new Gaussian proposal distribution to be used in conjunction with the sequential Monte Carlo (SMC) method for solving non-linear filtering problem. This proposal incorporates all the information about the to be estimated current state from both the available state and observation processes. This makes it more effective than the commonly used state transition density as a proposal, which ignores the recent observation. The introduced proposal is completely characterized by the exact moments obtained from the dynamical system. This is in contrast with recent works where the moments are approximated either numerically or by linearizing the observation model. Because of its Gaussian nature, it is also very easy to implement. We show further that the newly introduced proposal performs better than other similar proposal functions which also incorporate both state and observations.
    Original languageEnglish
    Pages (from-to)203-208
    Number of pages6
    JournalStatistics and computing
    Volume19
    Issue number2
    DOIs
    Publication statusPublished - 2009

    Keywords

    • Bayesian filtering
    • Nonlinear dynamic system
    • Sequential Monte Carlo methods
    • Particle filtering
    • Importance sampling
    • Moment matching

    Fingerprint Dive into the research topics of 'Gaussian proposal density using moment matching in SMC methods'. Together they form a unique fingerprint.

    Cite this