Robust Kalman smoother for EIV state space model based on multivariate least trimmed squares estimator

Abstract

This paper derives a robust Kalman smoother estimate for the errors-in-variables state space model that is less sensitive to outliers in the sense of the multivariate least trimmed squares (MLTS) method. Since the MLTS estimate is a combinatorial optimization problem, the randomized algorithm has been proposed. However, the uniform sampling method has a high computational cost and may lead to a biased estimate. Therefore, we apply the subsampling method. The algorithm presented here is both efficient and easy to implement. A Monte Carlo simulation result shows the efficiency of the proposed algorithm.