A New Robust Adaptive Fading Unscented Kalman Filter-Based Vector Tracking Loop for Mitigating GNSS NLOS Reception

Abstract

The positioning accuracy of global navigation satellite system (GNSS) in urban areas can be significantly affected by non-line-of-sight (NLOS) receptions. Vector tracking loop (VTL) which utilizes information from multiple satellites has been proposed for mitigating the NLOS detection. This paper develops a robust adaptive fading unscented Kalman filter (AF-UKF-) based navigation filter for NLOS mitigation in VTL-based systems. It integrates the AF-KF approach previously proposed for adaptive Bayesian KF to the UKF-based navigation filter to update the scale of the state and measurement noise covariances so that system noise and measurement uncertainties can be mitigated. A robust extension of the AF-UKF algorithm based on robust statistics is proposed to effectively detect and correct NLOS reception in each KF update. A robust initialization method is also proposed to provide a reliable warm start to the robust AF-UKF so that the latter can reliably start-up even under NLOS reception and limited initial position information. The proposed method was evaluated and compared with conventional algorithms using real GPS signal datasets and simulated data. Results showed that the proposed navigation solution yields more accurate and reliable positioning performance than conventional methods under NLOS reception. The advantage and stability of the AF-KF and AF-UKF are also analyzed.