State estimation with missing measurements using IMM

The missing of the measurements will deteriorate the estimation or even make the estimators divergent. We first analyze the performance of the existed estimation approaches in nonlinear systems with missing measurements. According to the effectiveness of the multiple model estimation over single model estimation, we propose to give the estimates of the states using the interacting multiple model estimation (IMM). The IMM contains two model sets. One of them corresponds to the systems modes and the other corresponds to the occurrence of the missing of the measurements. The simulation results show the proposed approach is more stable and accurate than the existed estimation approaches.

1 INTRODUCTION

The problem of state estimation is important in various applications ranging from target tracking and detection. Many useful estimation approaches have been proposed when there is no information loss in measurements. Multiple-model (MM) estimation approaches turned up to solve the problems which exist in single-model estimation and have been applied in many domains in practice. The main application involves air traffic control, ballistic missile defense and maneuvering target tracking [1] [2]. In MM algorithms, a set of models are set up to approximate the system modes. Interacting Multiple Model Algorithm (IMM) has been demonstrated to be the most cost-effective algorithm among all the MM estimation approaches.