Kullback-Leibler divergence -based improved particle filter

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

In this paper, we develop an improved particle filtering algorithm for nonlinear states estimation. In case of standard particle filter, the latest observation is not considered for the evaluation of the weights of the particles as the importance function is taken to be equal to the prior density function. This choice of importance sampling function simplifies the computation but can cause filtering divergence. In cases where the likelihood function is too narrow as compared to the prior function, very few particles will have significant weights. Hence a better proposal distribution that takes the latest observation into account is desired. The proposed algorithm consists of a particle filter based on minimizing the Kullback-Leibler divergence distance to generate the optimal importance proposal distribution. The proposed algorithm allows the particle filter to incorporate the latest observations into a prior updating scheme using the estimator of the posterior distribution that matches the true posterior more closely. In the comparative study, the state variables are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error with respect to the noise-free data. The simulation results show that the proposed algorithm, outperforms the standard particle filter, the unscented Kalman filter, and the extended Kalman filter algorithms.

Original languageEnglish
Title of host publication2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014
PublisherIEEE Computer Society
DOIs
Publication statusPublished - 2014
Event2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014 - Castelldefels-Barcelona, Spain
Duration: 11 Feb 201414 Feb 2014

Other

Other2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014
CountrySpain
CityCastelldefels-Barcelona
Period11/2/1414/2/14

Fingerprint

Importance sampling
Extended Kalman filters
State estimation
Kalman filters
Mean square error
Probability density function

Keywords

  • Kullback-Leibler divergence
  • Particle filter

ASJC Scopus subject areas

  • Signal Processing
  • Control and Systems Engineering

Cite this

Mansouri, M., Nounou, H., & Nounou, M. (2014). Kullback-Leibler divergence -based improved particle filter. In 2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014 [6808793] IEEE Computer Society. https://doi.org/10.1109/SSD.2014.6808793

Kullback-Leibler divergence -based improved particle filter. / Mansouri, Majdi; Nounou, Hazem; Nounou, Mohamed.

2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014. IEEE Computer Society, 2014. 6808793.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Mansouri, M, Nounou, H & Nounou, M 2014, Kullback-Leibler divergence -based improved particle filter. in 2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014., 6808793, IEEE Computer Society, 2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014, Castelldefels-Barcelona, Spain, 11/2/14. https://doi.org/10.1109/SSD.2014.6808793
Mansouri M, Nounou H, Nounou M. Kullback-Leibler divergence -based improved particle filter. In 2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014. IEEE Computer Society. 2014. 6808793 https://doi.org/10.1109/SSD.2014.6808793
Mansouri, Majdi ; Nounou, Hazem ; Nounou, Mohamed. / Kullback-Leibler divergence -based improved particle filter. 2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014. IEEE Computer Society, 2014.
@inproceedings{7f0674df1a8b4b3c8bf6e9d2af4b5765,
title = "Kullback-Leibler divergence -based improved particle filter",
abstract = "In this paper, we develop an improved particle filtering algorithm for nonlinear states estimation. In case of standard particle filter, the latest observation is not considered for the evaluation of the weights of the particles as the importance function is taken to be equal to the prior density function. This choice of importance sampling function simplifies the computation but can cause filtering divergence. In cases where the likelihood function is too narrow as compared to the prior function, very few particles will have significant weights. Hence a better proposal distribution that takes the latest observation into account is desired. The proposed algorithm consists of a particle filter based on minimizing the Kullback-Leibler divergence distance to generate the optimal importance proposal distribution. The proposed algorithm allows the particle filter to incorporate the latest observations into a prior updating scheme using the estimator of the posterior distribution that matches the true posterior more closely. In the comparative study, the state variables are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error with respect to the noise-free data. The simulation results show that the proposed algorithm, outperforms the standard particle filter, the unscented Kalman filter, and the extended Kalman filter algorithms.",
keywords = "Kullback-Leibler divergence, Particle filter",
author = "Majdi Mansouri and Hazem Nounou and Mohamed Nounou",
year = "2014",
doi = "10.1109/SSD.2014.6808793",
language = "English",
booktitle = "2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Kullback-Leibler divergence -based improved particle filter

AU - Mansouri, Majdi

AU - Nounou, Hazem

AU - Nounou, Mohamed

PY - 2014

Y1 - 2014

N2 - In this paper, we develop an improved particle filtering algorithm for nonlinear states estimation. In case of standard particle filter, the latest observation is not considered for the evaluation of the weights of the particles as the importance function is taken to be equal to the prior density function. This choice of importance sampling function simplifies the computation but can cause filtering divergence. In cases where the likelihood function is too narrow as compared to the prior function, very few particles will have significant weights. Hence a better proposal distribution that takes the latest observation into account is desired. The proposed algorithm consists of a particle filter based on minimizing the Kullback-Leibler divergence distance to generate the optimal importance proposal distribution. The proposed algorithm allows the particle filter to incorporate the latest observations into a prior updating scheme using the estimator of the posterior distribution that matches the true posterior more closely. In the comparative study, the state variables are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error with respect to the noise-free data. The simulation results show that the proposed algorithm, outperforms the standard particle filter, the unscented Kalman filter, and the extended Kalman filter algorithms.

AB - In this paper, we develop an improved particle filtering algorithm for nonlinear states estimation. In case of standard particle filter, the latest observation is not considered for the evaluation of the weights of the particles as the importance function is taken to be equal to the prior density function. This choice of importance sampling function simplifies the computation but can cause filtering divergence. In cases where the likelihood function is too narrow as compared to the prior function, very few particles will have significant weights. Hence a better proposal distribution that takes the latest observation into account is desired. The proposed algorithm consists of a particle filter based on minimizing the Kullback-Leibler divergence distance to generate the optimal importance proposal distribution. The proposed algorithm allows the particle filter to incorporate the latest observations into a prior updating scheme using the estimator of the posterior distribution that matches the true posterior more closely. In the comparative study, the state variables are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error with respect to the noise-free data. The simulation results show that the proposed algorithm, outperforms the standard particle filter, the unscented Kalman filter, and the extended Kalman filter algorithms.

KW - Kullback-Leibler divergence

KW - Particle filter

UR - http://www.scopus.com/inward/record.url?scp=84901290677&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901290677&partnerID=8YFLogxK

U2 - 10.1109/SSD.2014.6808793

DO - 10.1109/SSD.2014.6808793

M3 - Conference contribution

BT - 2014 IEEE 11th International Multi-Conference on Systems, Signals and Devices, SSD 2014

PB - IEEE Computer Society

ER -