GNSS precise positioning for smartphones based on the integration of factor graph optimization and solution separation

Abstract

Currently, the global navigation satellite system on smartphones suffers from frequent outliers (faults) in the measurements, limiting their precise positioning applications. To mitigate, the factor-graph optimization (FGO) method is used, where the Doppler and time-differenced carrier-phase (TDCP) measurements determine the velocities. However, their estimations are vulnerable since the Doppler measurements are inaccurate, while the TDCP measurements suffer from frequent cycle slips. Statistical testing methods can be used to exclude measurement outliers, but they have high volumes of false alarms and missed detections. In this study, we propose an improved scheme of FGO with a solution separation (SS) test to improve smartphone precise positioning. After excluding measurement outliers with the SS test for TDCP and time-differenced pseudorange observation models, we apply an improved FGO scheme for positioning. We conducted an experiment with Xiaomi Mi 8 and Google Pixel 5, showing that our method outperforms the conventional FGO and Kalman filtering methods.