Real-time multifrequency PPP–IAR performance with an integrated GPS/QZSS constellation

The use of overlapping frequencies in multi global navigation satellite system (multi–GNSS) constellations is an effective approach for accelerating the convergence of precise point positioning with integer ambiguity resolution (PPP–IAR). However, frequency-dependent biases related to receiver types exist between different constellations, making the correction of differential intersystem biases (DISB) essential for tight integration (TI). In this study, a TI GPS/QZSS PPP–IAR model is proposed that fully exploits the observations collected on overlapping frequencies provided by the QZSS to enable TI ambiguity resolution and ensure compatibility across different receiver types. Both loose integration (LI) and TI PPP–IAR experiments are conducted using GPS/QZSS data collected in China with two types of receivers: UB4b0 and P20. The experimental results reveal that the two types of receivers used in this study exhibit stable DISB between GPS and QZSS only at the L2 frequency. The results from both the server and user sides indicate that correcting the DISB can effectively restore the integer properties of intersatellite differential ambiguities between the GPS and the QZSS. With respect to the user-side positioning performance of PPP–IAR, the TI method outperforms the LI and GPS-only approaches in terms of convergence time, positioning accuracy. Further analyses of the fixed satellite number, PDOP, ratio, and ADOP metrics indicate that although TI exhibits satellite geometry similar to that of LI, it outperforms LI in terms of ambiguity resolution speed. Specifically, at the 95th percentile, the convergence time improvements of the TI approach over the LI approach are 47.5%, 34.1%, and 59.2% for the east, north, and up (E/N/U) components, respectively, and those over the GPS-only approach are 73.3%, 59.7%, and 64.6%, respectively. In terms of positioning accuracy, the TI approach achieves average improvements of 33.32%, 31.47%, and 28.42% over the GPS-only approach in the E/N/U components, respectively, and of approximately 10% over the LI approach in all components.