A New GNSS FDE Algorithm for Real-Time LEO Satellite Orbit Determination

As a kind of the aircraft with orbital altitudes between 400 and 2000 km, low earth orbit (LEO) satellites with the advantages of small size and low cost play an important role in various earth observation missions. To ensure the successful completion of each mission, it is essential to obtain precise and continuous orbit information of such satellites. Currently, global navigation satellite systems (GNSS) are mainly used for this task, but the complex electromagnetic environment in space renders on-board GNSS observation data unreliable, with serious implications for the accuracy of the orbit determination. This article therefore proposes a new GNSS fault detection and exclusion (FDE) algorithm for real-time LEO satellite orbit determination. Specifically, the advanced quasi-accurate detection (QUAD) based on orbit prediction model is proposed to detect GNSS anomalies according to the deterministic relationship between the errors and the observations. The validity of the proposed algorithm is proved by test results: in the case of a single GNSS satellite anomaly, the horizontal and vertical position accuracy of the proposed algorithm is improved by 71.6% and 83.2%, respectively, compared with the traditional w-test-based algorithm. As the number of anomalies increases, the proposed algorithm is shown to be able to circumvent up to four fault outliers, while keeping the 3-D position error lower than 10 m. Specifically, the probability of identifying outliers remains above 90%, and the 3-D position error is reduced by about 4 m in the worst case of dual anomalies, compared to the iterative w-test with observation separation algorithm.