An improved robust filtering against observation and satellite state anomalies in centralized autonomous orbit determination for GNSS

Abstract

Anomalies caused by the failure of ranging-related facilities and satellite orbit maneuvering will greatly affect the performance of Autonomous Orbit Determination (AOD) for Global Navigation Satellite System (GNSS). In view of that, we proposed an improved robust filtering named as Resisting Anomaly Robust Filtering (RARF) to improve the precision and enhance the reliability of AOD in situation of anomalies in observations and states of satellites (i.e. orbit maneuvering). We performed the centralized AOD with the RARF for a hybrid GNSS constellation using simulated observations, and analyzed its performance in cases of no anomaly, anomalies in observations only and anomalies in states of satellites only. The experimental results indicate that: (1) With anomalies in observations only, the RARF is much more robust than the extended Kalman filter (EKF), and results of AOD with the RARF are entirely free from abnormal observations; (2) In the situation of anomalies in states of satellites only, the precision of AOD with the RARF can reach to the order of 10 m after 1.5 h orbit recovery if an anchor is available. As the solving time extends, the precision of AOD can be up to dm level.