A VB-IMM filter for ADS-B data
Quanhui Wang, Jianjun Huang · 2014
A variational Bayesian approximation-based interacting multiple model (VB-IMM) filter for automatic dependent surveillance-broadcast (ADS-B) Data is proposed. ADS-B data is a type of measurements with unknown noise variances. The variational Bayesian adaptive Kalman filter (VB-AKF) is a recursively forming separable approximation to the joint distribution of both states and noise parameters by the variational Bayesian method. The proposed algorithm adopts the interacting multiple models (IMM) to update the iteration, and adjusts adaptively its parameters according to the accuracy category of measurements to solve the problem of maneuvering target tracking. Simulation results show that the proposed method can achieve better performance in tracking accuracy in the situations with unknown noise variance.