Robust real-time GNSS satellite clock estimation using atomic clock models with anomaly detection and mitigation

This is the authors' abstract. We don't add key points for this paper.

Real-time Precise Point Positioning (RT-PPP) relies heavily on precise Global Navigation Satellite System (GNSS) satellite clock products. Model-based clock estimation can exploit the improved stability of modern atomic clocks to establish and maintain a stable network timescale. However, conventional clock models cannot adequately handle clock anomalies, such as phase and frequency jumps, which can degrade clock estimation accuracy and compromise filter stability. To address this issue, we propose and validate an anomaly-aware real-time satellite clock estimation framework that integrates atomic clock modeling with anomaly identification, isolation, and recovery within a network solution, thereby limiting the propagation of clock anomalies into the network time reference. To further assess the reliability and practical applicability of the framework, the anomaly-confirmation parameters are systematically evaluated, and the computational efficiency is assessed to verify real-time feasibility. Subsequently, validation using actual GNSS data demonstrates the effective identification and mitigation of phase jumps, frequency jumps, and complex anomalies. Real-time service performance evaluation shows that the method achieves comparable or improved clock accuracy across multiple GNSS constellations, while significantly enhancing timescale stability. The proposed approach therefore provides a more stable and resilient real-time GNSS timescale, thereby supporting high-accuracy timing applications.