An efficient strategy for multi-GNSS real-time clock estimation based on the undifferenced method

Precise satellite clock product is an indispensable prerequisite for the real-time precise positioning service. To meet the requirement of numerous time-critical applications, real-time satellite clock corrections need to be broadcast to users with an update rate of 5 s or higher. With the rapid development of global navigation satellite systems (GNSS) over the past decades, abundant GNSS tracking stations and modern constellations have emerged, and the computation for multi-GNSS real-time clock estimation has become rather time-consuming. In this contribution, an efficient strategy is proposed to achieve high processing efficiency for multi-GNSS real-time clock estimation, wherein undifferenced method based on sequential least square is adopted. In the proposed strategy, parallel data processing and high-performance matrix operations are introduced to accelerate the processing of multi-GNSS clock estimation. The former is based on OpenMP (Open Multi-Processing), while the latter is achieved by the implementation of the Schur complement and the open-source library OpenBLAS. Multi-GNSS observations from 85 globally distributed tracking stations are employed for the generation of real-time precise clock products. The average elapsed time per epoch with the proposed strategy is 0.35, 0.68, and 2.30 s for GPS-only, dual-system, and quad-system solutions, respectively. Compared to the traditional serial strategy, the computation efficiency is significantly improved by 76.0%, 77.3%, and 77.7%, respectively. The accuracy of the estimated clocks is evaluated with respect to IGS final GPS clock products and GFZ final multi-GNSS clock products (GBM0MGXRAP), and multi-GNSS real-time precise point positioning (PPP) experiments are further carried out. All the results indicate that the proposed strategy is efficient, accurate, and can promise high-rate multi-GNSS real-time clock estimation.