Combining GPS and GLONASS in all-in-view for time transfer

GPS code measurements have been used for three decades for remote clock comparison, also called Time Transfer. Initially based on a technique using common-view (CV) single-frequency measurements, GPS time transfer now mostly uses dual-frequency measurements from geodetic receivers processed in all-in-view (AV). With the completion of the GLONASS constellation, it has been possible to readily use it in the CV single-frequency mode, providing results similar to GPS for short-distance time links. However GLONASS results are not readily equivalent to GPS in the dual-frequency AV mode, necessary for any moderate- to long-distance link, and this paper shows how to achieve this. We first present the GLONASS upgrade of the R2CGGTTS software, a tool to provide dual-frequency measurements in a format dedicated to time transfer named CGGTTS (Common GPS GLONASS Time Transfer Standard). The GLONASS navigation files are used to determine satellite clocks and positions, and dual-frequency pseudorange measurements are linearly combined to compute the CGGTTS results in a similar way as for GPS. In a second part, we present the combination of GPS and GLONASS into one unique time transfer solution based on AV. The results are first corrected using precise satellite orbit and clock products delivered by the IGS analysis centre ESOC, and characterized by the same reference for the GPS and GLONASS satellite clocks. Then, the need to introduce satellite-dependent hardware delays in GLONASS results is emphasized, and a procedure is proposed for their determination. The time transfer solutions obtained for GPS-only and GPS+GLONASS are then compared. The combination of GPS and GLONASS results in AV provides a time transfer solution having the same quality as GPS only. Furthermore, comparisons show that even when increasing the number of observations in CV thanks to the combination of the two constellations, the AV remains superior to the CV solution in terms of noise and short term stability, especially for long baselines.