Fast Clustering Satellite Selection Based on Doppler Positioning GDOP Lower Bound for LEO Constellation

With the rapid development of low Earth orbit (LEO) constellations, they have become an effective complement to the traditional global navigation satellite system (GNSS) in meeting positioning needs. However, the large number of available satellites and hardware limitations of the receivers make it impossible to fully apply the information from all visible satellites, hence it is necessary to quickly select a geometrically superior combination of LEO satellites for positioning. On the other hand, since the LEO satellites generally adopt multiple epochs positioning based on Doppler measurements due to their nonnavigation design, the geometric dilution of precision (GDOP), which is an important parameter for evaluating satellite combinations, and the common satellite selection methods in traditional GNSS cannot be applied directly. In this article, a Doppler-GDOP (DGDOP) model is developed for the LEO satellites, and a fast clustering satellite selection method for Doppler positioning (FCSDp) is proposed based on it. Experiments are performed using real Starlink signals, and the results show that using the FCSDp method proposed in this article to select satellite combinations for positioning could stably reduce the positioning error by more than 45% compared to the traditional method, which verifies the effectiveness of the method. Using the satellite selection method proposed in this article for LEO satellite positioning could ensure good positioning accuracy while satisfying low computational complexity.