IMPROVING GNSS POSITIONING RELIABILITY AND ACCURACY BASED ON FACTOR GRAPH OPTIMIZATION IN URBAN ENVIRONMENT

Abstract

Global navigation satellite system (GNSS) can provide global, precise, and continuous positioning in open-sky environments. However, urban environments with frequent outliers and cycle slips degrade the traditional Extended Kalman Filter (EKF) positioning performance. The susceptibility of EKF to outliers is attributed to its inherent structure. To mitigate, the GNSS positioning based on the Factor Graph Optimization (FGO) structure is adopted. FGO can enhance time correlation among observations and enable the updating of historical information, thereby improving resistance against outliers. In this study, we proposed a single-differenced GNSS-FGO model instead of the double-differenced model to preserve the sparsity of FGO, and outlier detection and PAR methods are employed to ensure urban positioning performance. To evaluate the proposed structure, experiments are conducted in both urban and open-sky environments. The results demonstrate the improvement of positioning accuracy and reliability, compared to EKF.