An algorithm for a near real-time data processing of GPS common-view observations

Abstract

Near real-time transfer of GPS common-view data is no longer a problem, but near real-time data processing of the data still calls for study, because it is not yet achieved by the usual smoothing and filtering techniques. Based on the characteristics of the GPS common-view data, a Kalman filtering algorithm is designed for estimating the time difference between two sites, while greatly reducing the observational noise. The algorithm is applied to the time difference between the National Time Service Center (NTSC) of China and the Communications Research Laboratory (CRL) of Japan (Over 2000 km apart), and to that between the CRL and the Korean Research Institute of Standards and Sciences (KRIS) (over 1000 km apart). The root mean square errors of the results obtained by the Kalman filter relative to those obtained from the Circular T of BIPM are less than 2.9 ns and 2.6 ns, in the two cases. Further, it is pointed out that, when multi-site data within a common-view network are available we can further improve the accuracy of the time comparisons by indirect observation adjustment. This statement was justified by application to the data from all three stations, i.e. NTSC, CRL, and KRIS.