Stochastic Modeling for Real-Time Kinematic GPS/GLONASS Positioning

Abstract

It is well known that for satellite-based high-precision kinematic positioning, the correct integer ambiguities must be identified on-the-fly (OTF). It has also been noted that reliable integer ambiguity resolution is highly dependent on applying correct stochastic models for differenced GPS and GLONASS measurements. Stochastic modeling for real-time kinematic (RTK) positioning, however, is a difficult task to accomplish. In this study, a practical method is proposed for directly estimating the variance and covariance components for the differenced GPS and GLONASS measurements. The applicability of the proposed method for RTK positioning has been tested with both GPS dual-frequency and combined GPS/GLONASS single-frequency datasets. Test results show that using the estimated measurement covariance matrices significantly improves the success rates of ambiguity resolution and the accuracy of positioning results.