The increasing number of GPS receivers for spaceborne applications has revitalized the interest in real-time and on-board orbit determination to increase the spacecraft autonomy and reduce the required amount of ground operations. Aside from high-precision applications that require a direct processing of raw code and phase measurements, the orbit determination can be based on the navigation solution generated by most of the GPS receivers. In general position values are provided with a spherical 1 σ accuracy of about 100 m, whereas velocity is only accurate to 1 m/s in representative spaceborne receivers. Using a KaIman filter and an appropriate dynamical model, the inherent measurement noise may be reduced considerably and much more accurate state vectors be obtained.
The filtering of the SPS (Standard Positioning Service) data provides various benefits for on-board navigation. These are mainly related to the fact that the resuIting trajectory is constrained by the dynamical laws of motion, whereas subsequent GPS position solutions are essentially uncorrelated. As such, the adjusted orbit information is less sensitive to the intentional deterioration of the GPS measurements (Selective Availability) than the navigation solution itself. In addition, the dynamical modeling allows a smooth interpolation of the trajectory, a bridging of data gaps and a detection of erroneous measurements. Finally, the KaIman filter provides precise velocity information as part of the estimated state vector, which results from the accumulation of position knowledge over extended data ares.
O. Montenbruck, E. Gill, "Satellite Orbits: Models, Methods and Applications", Springer Verlag, Heidelberg; 2005.