Nonlinear filtering in unknown measurement noise and target tracking system by variational Bayesian inference

作者：Yu, Xingkai; Li, Jianxun^*; Xu, Jian

来源：Aerospace Science and Technology, 2019, 84: 37-55.

DOI：10.1016/j.ast.2018.08.043

• Nonlinear filtering
• Target tracking system
• Probability density function
• Variational Bayesian inference
• Iterative algorithm
• Unknown measurement noise

This paper considers a class of nonlinear filtering algorithms based on variational Bayesian (VB) theory to settle the unknown measurement noise problem in target tracking system. When the unknown measurement noise is conditionally independent of states, based on the variational idea, estimate of probability density function of state is converted into approximation two probability density functions for both unknown noise and nonlinear states. Then, an iterative algorithm is established to jointly estimate the state and the unknown measurement noise using variational Bayesian inference. Thus, the unknown measurement noise could be estimated as hidden state. The convergence result of the proposed nonlinear probability density function approximation algorithm is also given. The simulation results of typical examples show that the proposed VB based methods have superior performance to these classic algorithms in target tracking problems.

• 出版日期2019-1
• 单位中国科学院电工研究所; 上海交通大学