A method of real-time long-baseline time transfer based on the PPP-RTK

Abstract

Long-baseline time transfer can nowadays reach rather high frequency stability based on post-processed batch least-squares adjustment using the Precise Point Positioning (PPP) or Integer-PPP (IPPP) methods. For real-time PPP users, time transfer results are degraded due to the filter-based processing mode, and the degraded accuracy of the real-time satellite orbits and clocks compared to the final ones. The Real-Time Kinematic (RTK) time transfer can significantly reduce the satellite-related errors, but has limits on the baseline length similar to the RTK positioning. Also, the delivery of raw observations instead of State-Space Representation (SSR) products could result in pressure on data transfer and difficulties related to latency and prediction. In this study, the PPP-RTK technique, which combines the advantages of the PPP and the RTK methods, is tested for real-time long-baseline time transfer. As an alternative approach to the above two methods, it allows for the time transfer of long baselines, while not relying on external high-sampling and high-precision satellite clocks. By delivering the satellite clocks and satellite phase biases produced within the PPP-RTK regional network, time differences can be estimated between users and the reference network station, with which stable time transfer between users separated by long baselines can be realized. Using dual-frequency GPS and Galileo data, the PPP-RTK time transfer is tested using approximately a thousand-kilometer-scale network in Europe. The time transfer results between two hydrogen masers, i.e., those on the 884 km baseline BRUX-ONSA and the 920 km baseline WTZR-ONSA, are computed. At an averaging time of 105 s, Modified Allan Deviation (MDEV) at the level of sub-10-15 to 10-15 can be reached when processing the user coordinates in the station fixed, static, or kinematic modes. The median clock residuals can converge to 1 ns and 0.3 ns within 2 min and 15 min, respectively, in the kinematic mode, while in the static and fixed modes the convergence times are shorter. With the augmentation of 150 Low Earth Orbit (LEO) satellites having simulated observations, the clock residuals can converge to 1 ns and 0.3 ns within 30 s and 3.5 min, respectively, for all the three estimation modes.