An analysis of intersystem biases for multi-GNSS positioning

Abstract

The capability of commercial receivers to track multiple GNSSs poses the problem of mutual alignments of reference frames and time scales. System biases should possibly be computed in advance and made known to the user, so that the only unknowns for code point positioning are the receiver coordinates and the receiver clock offset relative to a unique time scale. We have examined data from the five GNSS constellations GPS, GLONASS, Galileo, QZSS and BeiDou for 4 days in 2013, nine European stations and four different GNSS receivers, with the goals to (a) evaluate the quality of the broadcast ephemeris relative to the precise orbits and clocks, when available (GPS, GLONASS, Galileo, QZSS), (b) estimate systematic offsets of the time scales kept by each GNSS relative to GPS and (c) investigate possible receiver-dependent system biases. We report an agreement at the meter level between coordinates of satellites computed with broadcast ephemeris and precise orbits. Differences between broadcast and precise clocks are up to ± few nanoseconds (1 ns = 10?9 s) for GPS (IGS products). For GLONASS, the difference of the broadcast clock to the Russian (IAC) SP3 values is also smaller than ±10 ns, but rises to about 200 ns when the broadcast clock is compared to the combined IGL product of IGS. For QZSS, the difference broadcast to SP3 products of the JAXA are also confined to ±10 ns. For Galileo, for which data are broadcast in a test mode, we report different values depending on the F-NAV or I-NAV satellite clock model: Using the SP3 products of Technical University of Munich (TUM), we report discrepancies between broadcast and SP3 of some tens of nanoseconds. System Time Biases lump together satellite-dependent Differential Code Biases, receiver-dependent Inter System Biases and offsets of a given GNSS time scale to a reference (we assume GPS) time scale. The System Time Bias is typically of the order of 360---380 ns for GLONASS, quite consistently with broadcast or precise (IAC) ephemeris. For Galileo, the biases are more variable and depend on the use of broadcast or precise data. For QZSS, we evaluate a QZSS to GPS System Time Bias very nearly constant in time. Its value differs of about 20 ns depending on the SP3 data source (TUM or JAXA). For BeiDou, we find Time System Biases of the order of 10---100 ns depending on day and receiver type. Receiver-dependent Time System Biases relative to GPS are quite evident in several different stations and are reasonably consistent across receiver type. We conclude by emphasizing the need for a monitoring activity of such inter-GNSS biases, both GNSS dependent and receiver dependent, and that appropriate means are devised for making them available to the final user.