Outlier detection based on variance estimation for satellite clock offset and its application in Galileo HAS service

Abstract Outlier detection and mitigation in satellite clock offsets are prerequisites for ensuring high-precision positioning and timing services. This paper proposes a quartile-based variance estimation method dedicated to satellite clock offset outlier detection. A systematic theoretical analysis and robustness evaluation, based on the breakdown point and influence function, are conducted and compared with the commonly used median-based and Allan variance methods. Simulation experiments reveal that the quartile-based method achieves superior accuracy with higher computational efficiency under low anomaly contamination, while the median-based method is recommended for high-anomaly scenarios. The real-time clock offset data of high accuracy service (HAS) from Galileo are used to validate the outlier detection performance of the methods, and the results show both quartile-based and median-based methods achieve rapid and stable convergence, significantly outperforming sample-based and Allan variance approaches. Precise Point Positioning (PPP) experiments further confirm that the accuracy of horizontal PPP has significantly improved in the N-direction and E-direction, with enhancement magnitudes reaching 10%-30% in some cases based on robust variance estimation methods. However, in the U-direction, the improvement in accuracy is limited.