Kullback-Leibler divergence for interacting multiple model estimation with random matrices

TL;DR

This paper introduces a novel IMM estimation method for jump Markov linear systems with unknown measurement noise covariance, utilizing Kullback-Leibler divergence to improve estimation accuracy.

Contribution

It proposes a new approach to combine inverse-Wishart distributions in IMM estimation by minimizing Kullback-Leibler divergence, providing a closed-form solution.

Findings

The proposed filter outperforms previous moment matching methods.

Simulation results demonstrate improved estimation accuracy.

The method effectively handles unknown measurement noise covariance.

Abstract

This paper studies the problem of interacting multiple model (IMM) estimation for jump Markov linear systems with unknown measurement noise covariance. The system state and the unknown covariance are jointly estimated in the framework of Bayesian estimation, where the unknown covariance is modeled as a random matrix according to an inverse-Wishart distribution. For the IMM estimation with random matrices, one difficulty encountered is the combination of a set of weighted inverse-Wishart distributions. Instead of using the moment matching approach, this difficulty is overcome by minimizing the weighted Kullback-Leibler divergence for inverse-Wishart distributions. It is shown that a closed form solution can be derived for the optimization problem and the resulting solution coincides with an inverse-Wishart distribution. Simulation results show that the proposed filter performs better than the previous work using the moment matching approach.