Analysis of long-term distributed autonomous orbit determination for BeiDou-3 satellites

With the support of inter-satellite link technology, GNSS can theoretically achieve the distributed autonomous orbit determination (AOD) function. Traditional AOD operation generally utilizes the forecast ephemeris uploaded by operational control segment (OCS) as the filter reference orbits or to constrain the orbit systematic errors, especially for constellation overall rotation effects in Earth-centered inertial (ECI) coordinate system. To get rid of the dependency on forecast trajectories for saving the OCS workload and also reduce the onboard storage and computation burden, we use a sequential extended Kalman filter to estimate the orbit parameters and consider main perturbation forces acting on satellites in the AOD solution. In particular, for modeling solar radiation pressure (SRP), an empirical prediction function derived by historical SRP estimates is introduced. Using the proposed scheme, the orbit 3D accuracy and user range error (URE) of the first 180-day distributed AOD solution for BeiDou-3 MEOs with precise Earth rotation parameters (ERPs) can reach about 2.10 and 0.43 m, respectively. The constellation rotation errors implied in AOD orbits around the X-, Y- and Z-axis of ECI system are less than 15.0, 11.7 and 15.2 mas, respectively. For real-world AOD scenarios, precise ERP is not available for satellites. With the 180-day prediction ERP, the orbit 3D errors and URE due to the gradually increased UT1-UTC error can be elevated to 14.62 and 2.91 m during our AOD experiments. Result analysis shows if OCS can upload latest prediction ERP at a frequency of once a week, the 180-day distributed AOD is expected to consistently produce real-time orbits preferable to broadcast ephemeris derived by the traditional region L-band tracking network.