Extended Kalman filter under maximum correntropy criterion
Authors:
- 3 November 2016
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1733-1737
Abstract
As a nonlinear extension of Kalman filter, the extended Kalman filter (EKF) is also based on the minimum mean square error (MMSE) criterion. In general, the EKF performs well in Gaussian noises. But its performance may deteriorate substantially when the system is disturbed by heavy-tailed impulsive noises. In order to improve the robustness of EKF against impulsive noises, a new filter for nonlinear systems is proposed in this paper, namely the maximum correntropy extended Kalman filter (MCEKF), which adopts the maximum correntropy criterion (MCC) as the optimization criterion instead of using the MMSE. In MCEKF, the state mean and covariance matrix propagation equation are used to obtain a prior estimation of the state and covariance matrix, and then a fixed-point algorithm is used to update the posterior estimates. The robustness of the new filter is confirmed by simulation results.Keywords
This publication has 23 references indexed in Scilit:
- Kernel recursive maximum correntropySignal Processing, 2015
- Convergence of a Fixed-Point Algorithm under Maximum Correntropy CriterionIEEE Signal Processing Letters, 2015
- Steady-State Mean-Square Error Analysis for Adaptive Filtering under the Maximum Correntropy CriterionIEEE Signal Processing Letters, 2014
- Hidden state estimation using the Correntropy Filter with fixed point update and adaptive kernel size2012
- Adaptive background estimation using an information theoretic cost for hidden state estimation2011
- Kernel adaptive filtering with maximum correntropy criterion2011
- Correntropy as a novel measure for nonlinearity testsSignal Processing, 2009
- The extended Kalman filter as an exponential observer for nonlinear systemsIEEE Transactions on Signal Processing, 1999
- The Kalman filter: A robust estimator for some classes of linear quadratic problemsIEEE Transactions on Information Theory, 1976
- A New Approach to Linear Filtering and Prediction ProblemsJournal of Basic Engineering, 1960