Real-Time GPS Satellite Clock Estimation Based on OpenMP

Abstract

The undifferenced method is the usual way to estimate GPS satellite clock corrections by dealing with globally distributed reference stations. However, the task is very time consuming when estimating a lot of ambiguity parameters, so it is difficult to generate high-rate real-time GPS clock corrections. In order to obtain 1 Hz GPS clock corrections based on intensive reference stations, the analysis center of IGS, namely GFZ, CODE uses the epoch difference approach to remove ambiguity parameters. And yet, the satellite orbit must be known in advance, because the models for orbit determination and for clock corrections estimation are different. According to this paper, the most time-consuming two parts are gain matrix computation and covariance matrix update. Therefore, the single observable Kalman filter is involved directly to speed up gain matrix computation and OpenMP technique is also used to accelerate the covariance matrix update. In order to analyze the accuracy of GPS satellite clock corrections and computing efficiency of GPS satellite clock estimation, the global 49 IGS reference stations distributed uniformly are chosen to compute satellite clock corrections between 2015-10-7 and 2015-10-7 and interval of estimated clock corrections is 60 s. In experiment, the result shows that using 16 cores in server to estimate satellite clock needs 1 s every epoch and compared to IGS 30 s clock products, the difference is smaller than 0.5 ns. So, the approach proposed in this paper can satisfy the 1 Hz GPS satellite clock real-time estimation.