Clock modelling techniques for an enhanced GNSS orbit determination

The determination of GNSS orbits is generally based on the processing of pseudorange and carrier phase measurements from a station network, with an Orbit Determination and Time Synchronization (ODTS) process. This process involves the satellite and ground station clocks as part of the GNSS measurement reconstruction. The clocks are generally estimated as a snapshot parameter, without assuming any correlation between epochs. However, the stability of satellite and some station clocks, based on technologies of hydrogen, cesium or rubidium, allows for a significant predictability. Taking advantage of this predictability the ODTS process can be improved, especially in those cases where the station network is limited or does not provide a good coverage for certain areas.

The clock modelling can be directly done by estimating additional parameters in the filter. A quadratic model is generally estimated for each clock, keeping a small snapshot contribution to account for the stochastic part and for potential deviations with respect to the theoretical behavior of the clock. The detection of this kind of deviations in the satellite and station clocks becomes a major factor for achieving a good performance with these techniques. In case the clock experiences feared events like phase or frequency jumps, the estimated clock model stops being valid and the estimation of model parameters needs to be reset.

In case a composite clock algorithm is used to provide the reference timescale for the ODTS, the estimation of clock models can rely on this algorithm. Algorithms of composite clock are generally based on a Kalman filter that estimates as part of the state vector the differences between each contributing clock and the composite timescale. These differences can be used not only to define the reference timescale of the ODTS, but also to remove the deterministic part of the clocks in the measurement reconstruction. As for the case of clock modelling, for algorithms of composite clock the detection and correction of anomalies in the contributing clocks becomes a critical point.

In this work, the integration of orbit determination, clock modelling and composite clock algorithms will be described. The impact of clock modeling techniques on the GNSS orbit determination accuracy will be presented, both considering a direct estimation of clock models in the ODTS and the estimation provided by the composite clock algorithm. These analyses will be based on NEODIS, the orbit determination software developed by Thales Alenia Space, which integrates with a Kalman filter approach GNSS orbit determination and composite clock algorithms.