中国邮电高校学报(英文) ›› 2010, Vol. 17 ›› Issue (4): 47-51.doi: 10.1016/S1005-8885(09)60486-6

• Wireless • 上一篇    下一篇

Tracking application about singer model based on marginalized particle filter

周非1,何伟俊2   

  1. 1. 重庆邮电大学无线定位与空间测量研究所
    2. 重庆邮电大学通信学院
  • 收稿日期:2009-12-18 修回日期:2010-05-06 出版日期:2010-08-30 发布日期:2010-08-31
  • 通讯作者: 何伟俊 E-mail:zhoufei@cqupt.edu.cn
  • 基金资助:

    This work was supported by the Science Project of Chongqing Educational Committee (KJ080520), and the Natural Science Foundation of Chongqing CSTC (CSTC, 2008BB2412).

Tracking application about singer model based on marginalized particle filter

ZHOU Fei,HE Wei-jun, FAN Xin-yue   

  1. Institute of Wireless Location and Space Measurement, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • Received:2009-12-18 Revised:2010-05-06 Online:2010-08-30 Published:2010-08-31
  • Contact: He Wei-Jun E-mail:zhoufei@cqupt.edu.cn
  • Supported by:

    This work was supported by the Science Project of Chongqing Educational Committee (KJ080520), and the Natural Science Foundation of Chongqing CSTC (CSTC, 2008BB2412).

摘要:

This article deals with the problem of maneuvering target tracking which results in a mixed linear/non-linear model estimation problem. For maneuvering tracking system, extended Kalman filter (EKF) or particle filter (PF) is traditionally used to estimate the states. In this article, marginalized particle filter (MPF) is presented for application in a mixed linear/non-linear model estimation problem. MPF is a combination of Kalman filter (KF) and PF. So it holds both advantage of them and can be used for mixed linear/non-linear substructure, where the conditionally linear states are estimated using KF and the nonlinear states are estimated using PF. Simulation results show that MPF guarantees the estimation accuracy and alleviates the potential computational burden problem compared with PF and EKF in maneuvering target tracking application.

关键词:

marginalized particle filter, Kalman filter, particle filter, maneuvering target tracking

Abstract:

This article deals with the problem of maneuvering target tracking which results in a mixed linear/non-linear model estimation problem. For maneuvering tracking system, extended Kalman filter (EKF) or particle filter (PF) is traditionally used to estimate the states. In this article, marginalized particle filter (MPF) is presented for application in a mixed linear/non-linear model estimation problem. MPF is a combination of Kalman filter (KF) and PF. So it holds both advantage of them and can be used for mixed linear/non-linear substructure, where the conditionally linear states are estimated using KF and the nonlinear states are estimated using PF. Simulation results show that MPF guarantees the estimation accuracy and alleviates the potential computational burden problem compared with PF and EKF in maneuvering target tracking application.

Key words:

marginalized particle filter, Kalman filter, particle filter, maneuvering target tracking