An intelligent joint filter for vector tracking loop considering noise interference

Abstract

In this paper, we propose an intelligent joint filter (JF) for enhancing the performance of vector tracking loop (VTL) in the Global Navigation Satellite System (GNSS). The JF combines the advantages of extended Kalman filter (EKF) and unbiased finite-impulse response (UFIR) filter. To this end, a supervised machine learning algorithm, named Gaussian mixture model (GMM) clustering, was used for providing excellent joint strategy. Those three types of filter-based vector tracking loop were first implemented and then processed with a set of raw satellite signals based on the software-defined receiver (SDR). Finally, comparative analyses and results of the tracking performance of EKF/UFIR/JF were carried out. Results show that the EKF-VTL has optimal tracking performance but sensitive to the noise statistics, which means it's not robust. The UFIR-VTL is suboptimal but more robust compare to EKF-VTL. The proposed JF-VTL is both optimal and robust.