A BDS/5G hybrid localization algorithm based on adaptive variational Bayesian for UAV positioning

Abstract

Due to the fact that the BeiDou navigation satellite system (BDS) signal is easily blocked when the unmanned air vehicle (UAV) shuttles between urban buildings, the positioning accuracy is limited or the positioning cannot be completed. Therefore, the 5th generation mobile communication technology (5G) positioning is introduced to establish a hybrid positioning system of BDS pseudo-range combined with 5G time of arrival (TOA) and angle of arrival (AOA). The noise distribution of the observation data has strong randomness, which leads to the contamination of the update of the prior covariance matrix in the Kalman filter (KF) prediction step, and affects the optimal estimation of the UAV position. Therefore, an adaptive variational Bayesian (VB) localization algorithm is proposed. The algorithm first uses the least squares (LS) solution of the positioning observation as the observation input of the KF, and judges the distribution type of the original observation noise according to the Grubbs criterion. Then, the VB update factor of the covariance matrix is adaptively adjusted according to the Gaussian or heavy-tailed non-Gaussian noise distribution to optimize the position estimation. The simulation results show that the proposed algorithm can achieve high-precision positioning and anti-interference performance under different states of UAV, different degrees of satellite occlusion, and different probability of random interference.