Jianghui Geng, Jiang Guo, Xiaolin Meng, Kefu Gao

Precise point positioning (PPP) has been suffering from slow convergences to ambiguity-fixed solutions. It is expected that this situation can be relieved or even resolved using triple-frequency GNSS data. We therefore attempt an approach where uncombined triple-frequency GPS/BeiDou/Galileo/QZSS (Quasi-zenith satellite system) data are injected into PPP, whereas their raw ambiguities are mapped into the extra-wide-lane, wide-lane and narrow-lane combinations for integer-cycle resolution at a single station (i.e., PPP-AR). Once both extra-wide-lane and wide-lane ambiguities are fixed to integers, the resulting unambiguous (extra-)wide-lane carrier-phase can usually outweigh the raw pseudorange to improve the convergence of positions and narrow-lane ambiguities. We used 31~days of triple-frequency multi-GNSS data from 76 stations over the Asia-Oceania regions and divided them into hourly pieces for real-time PPP-AR. We found that the positioning accuracy for the first ten minutes of epochs could be improved by about 50\% from 0.23, 0.18 and 0.43~m to 0.12, 0.08 and 0.27~m for the east, north and up components, respectively, once wide-lane ambiguity fixing was achieved for triple-frequency PPP. Consequently, 48\% of PPP solutions could be initialized successfully with narrow-lane ambiguities resolved within 2~minutes, in contrast to only 26\% for dual-frequency PPP. On average, 6~minutes of epochs were required to achieve triple-frequency PPP-AR whereas 9~minutes for its dual-frequency counterpart. Of particular note, the more satellites contribute to triple-frequency PPP-AR, the faster the initializations will be; as a typical example, the mean initialization time declined to 3~minutes in case of 20--21 satellites. We therefore envision that only a few minutes of epochs will suffice to reliably initialize real-time PPP once all GPS, BeiDou, Galileo and QZSS constellations emitting triple-frequency signals are complete in the near future.