A New Criterion Construction and Verification for GNSS Satellite Selection Based on Near-Real-Time Accuracy

Global Navigation Satellite Systems (GNSS) have undergone more than half a century of development and construction, with more than a hundred navigation satellites currently providing precise and reliable positioning, navigation, and timing (PNT) services for various users. Meanwhile, efficient utilization of these satellites has become a topic of interest. Selecting an appropriate satellite set in a proper manner can reduce computational burden while ensuring positioning accuracy. Geometric Dilution of Precision (GDOP) is commonly used in satellite selection as it quantifies the impact of satellite geometry on positioning accuracy. Due to its computational simplicity, GDOP has been widely applied in satellite selection, but it only considers the satellite geometric configuration while ignoring the quality of satellite observations. As a result, the selected satellite set may lead to poor positioning accuracy. To address this issue, we use a satellite selection criterion based on the combination of near-real-time accuracy of satellite observations and geometric configuration. This criterion utilizes the combination of Geometry-Free Ionosphere-Free (GFIF) and Melbourne-Wübbena (MW) linear combinations of observations. Through a sliding window, we estimate the near-real-time accuracy of observations and use it to calculate the Weighted Geometric Dilution of Precision (WGDOP) for satellite selection. In a global International GNSS Service (IGS) station validation experiment, the satellite set selected based on WGDOP using near-real-time accuracy of GFIF and MW observations improved overall positioning accuracy by 11.6% and 12% when compared with the GDOP-based selection, and by 6% and 6.4% when compared with the Signal-to-Noise Ratio (SNR) weighting method. In a low-cost device validation experiment, the satellite selection method based on near-real-time accuracy of GFIF and MW improved positioning accuracy by 22.5% and 19.7% when compared with the GDOP-based method, and by 23.3% and 20.5% when compared with the SNR-based method. A set of dynamic observation experiments further demonstrates that the satellite selection method based on the near-real-time accuracy of GFIF and MW combinations outperforms the other two selection criteria in dynamic scenarios.