An ensemble Kalman filtering algorithm for state estimation of jump Markov systems

Abstract

This paper presents a target model whose dynamics is modelled as a jump Markov process. The problem of state estimation of jump Markov system using ensemble Kalman filter (EnKF) is dealt here. The EnKF is designed for higher order nonlinear state estimation. In target tracking using jump process, accuracy of the state estimate is defined as a function of ensemble size. The jump Markov system minimises state error at the state estimation and assumes that all probability distributions involved are to be Gaussian. This paper proposes state estimation of jump Markov systems using EnKF, at each step the ensemble of state estimates are calculated. From the result, calculate the error statistics from ensemble forecasts which are dynamically propagated through nonlinear system.