Differential Code Bias Estimation using Multi-GNSS Observations and Global Ionopshere Maps

Measurements of Global Navigation Satellite System (GNSS) receivers are affected by systematic offsets related to group and phase delays of the signal generation and processing chain. The resulting code and phase biases depend on the transmission frequency and the employed signal modulation. Within this study differential code biases (DCBs) of legacy and modernized GNSS signals are derived from pseudodrange observations of a global multi-GNSS receiver network. Global ionosphere maps (GIMs) are employed for the correction of ionospheric path delays. Satellite and receiver-specific contributions are separated based on the assumption of additive biases and a zero-mean condition for the satellite biases within a constellation. Based on 6 months of data collected within the Multi-GNSS Experiment (MGEX) of the International GNSS Service (IGS), DCBs for the publicly available signals of GPS, Galileo and BeiDou have been determined. The quality of the resulting DCB estimates is assessed and compared against group delay parameters transmitted by the GNSS providers as part of the broadcast ephemeris data. INTRODUCTION Pseudorange observations are well known to be affected by signaland frequency-dependent differential code biases (DCBs) that need to be considered in the observation modeling. Depending on the choice of a conventional signal or signal combination for the clock offset determination, timing group delays (TGDs) or inter-signal corrections (ISCs) need to be applied when using other signals for pseudorange-based positioning. DCBs and TGDs are routinely determined for the legacy GPS and GLONASS signals (C/A and P/Y code on L1 and L2) but only limited knowledge is presently available for satellite and receiver DCBs related to modernized GPS signals (L2C, L5) as well as those of new and emerging constellations (BeiDou, Galileo, QZSS, IRNSS). Within the Multi-GNSS Experiment (MGEX) of the International GNSS Service (IGS) multi-frequency observations and broadcast navigation messages of the BeiDou and Galileo constellations as well as the QZS-1 satellite are collected on a routine basis by a global network of monitoring stations. These offer a basis for an independent determination of differential code biases for a wide range of signal combinations within a combined estimation of DCBs and ionospheric parameters. As a simplified alternative, a multi-GNSS DCB estimation system has been established, which makes use of global ionosphere maps (GIMs) to model the contribution of ionospheric path delays on the difference of dualfrequency pseudoranges. DCBs of individual satellitereceiver combinations can thus be derived from an average of ionosphere-corrected pseudoranges differences over a given tracking arc for any combination of commonly tracked signals. Using a distributed network of monitoring stations and observations of all satellites within a constellation, the satellite-receiver DCBs may subsequently be partitioned into satellite-specific and receiver-specific DCBs. In accord with the common convention applied for GPS and GLONASS within the IGS, a zero-mean condition is applied for the constellation average of the satellite biases in this process. Following a brief overview of the MGEX tracking network and the employed receiver types, the concept of DCB estimation with a priori ionosphere information is presented. Thereafter, results for legacy and modernized GPS signals as well as Galileo and BeiDou are presented and discussed. To assess the achievable accuracy, DCBs for legacy GPS signals are compared against DCBs derived routinely by the Center for Orbit Determination in Europe (CODE) from the much larger core IGS network. Furthermore, DCBs for new constellations are compared against broadcast group delay parameters determined by the GNSS providers. TRACKING NETWORK AND RECEIVERS In response to the ongoing modernization of legacy GNSSs (GPS, GLONASS) and the rapid build-up of multiple new satellite navigation systems (BeiDou, Galileo, QZSS, IRNSS), the International GNSS Service (IGS) has established the Multi-GNSS Experiment (MGEX; [1]) as a platform for early familiarization with new systems and signals. Starting in early 2012 a new network of multiGNSS receivers was established in parallel to established GPS/GLONASS stations. By the end of 2013, the MGEX network has already grown to roughly 90 stations (Figure 1) and offered a global coverage for Galileo and BeiDou satellites in medium-altitude Earth orbits (MEOs) as well as regional coverage of the QZSS and BeiDou satellites in inclined geosychronous and geostationary orbits (IGSs/GEOs). Figure 1 MGEX station distribution and supported constellations (Dec. 2013) Other than various private multi-GNSS networks deployed by companies such as Trimble or Fugro, the MGEX network is highly heterogenous. It makes use of a wide variety of receivers and antennas as well as diverse combinations thereof. An up-to-date overview of the MGEX network and the employed hardware is available through the MGEX website [2]. For the period covered by this study (Jan.-June 2013), the MGEX network had not reached its current deployment status and was therefore augmented by stations of the COperative Network for GNSS Observations (CONGO) that joined the MGEX network at a later stage. A summary of contributing receivers and their observation types is provided in Table 1. Overall, GPS, Galileo, and BeiDou observations from up to 85 receivers were incorporated into the DCB analysis. While the majority of these receivers also tracks the (legacy) GLONASS signals (C/Aand P-code on L1 and L2), the estimation of GLONASS DCBs is already well covered by the IGS and beyond the scope of the present study. Likewise, QZSS has been excluded from the analysis, since the availability of only a single spacecraft does not enable a proper separation of satellite and receiver DCBs without a calibrated reference station. Table 1 Receiver and observation types used for the estimation of differential code biases. Observation types for GPS (G) Galileo (E), BeiDou (C) are based on RINEX 3 observation codes [3]. Receiver Type Sites Observations Javad TR_G2T, TRE_G3TH 29 G: 1C,1W,2X,2W,5X E: 1X,5X Javad TRE_G3TH (v8 board) 1 G: 1C,1W,2X,2W,5X E: 1X,5X,7X,8X C: 2I,7I Trimble NETR9 29 G:1C,2X,2W,5X E: 1X,5X,7X,8X C: 2I,6I,7I Leica GR10, GR25, GRX1200+GNSS 14 G: 1C,2S,2W,5Q E: 1C,5C,7C,8Q NovAtel OEM6 1 G: 1C,2W,5Q E: 1C,5Q Septentrio AsteRx3, PolaRxS/4/4TR 11 G: 1C,2L,2W,5Q E: 1C,5Q,7Q,8Q C: 2I,7I Most modernized GNSS signals offer distinct data-less (pilot) signal components in parallel to those modulated with navigation data. These pilot signals are considered to facilitate a robust signal tracking under adverse conditions, since the coherent integration time is not limited by bit transitions. Inspection of Table 1 shows that the currently available multi-GNSS receivers can largely be divided into two categories with respect to tracking of such signals. While part of the receivers (Septentrio, NovAtel, Leica) provide pilot-only observations, others (Javad, Trimble) provide measurements from a combined pilot+data tracking. In case of the GPS L2C signal, a total of three flavors can even be encountered: observations derived from tracking of the medium length L2C(M) code with navigation data (Leica), the long L2C(L) pilot component (Septentrio) or the combined L2C(L+M) signal (Javad, Trimble). Other than specialized Test User Receivers [4] which may support concurrent (or at least configurable) tracking and measurement generation for multiple signal components, only a single, predefined tracking mode is usually available in commercial multi-GNSS receivers. For the determination of differential code biases it is, nevertheless, important to carefully distinguish different tracking modes, since the satellite contribution to the observed receiver-plus-satellite DCB may depend on the particular signal component(s) selected for the tracking. In accord with this consideration, distinct group delay corrections will be broadcast for the Iand Q-components of the GPS (and QZSS) L5 and L1C signals ([5]-[6]). For L2C, only a single Inter-Signal Correction (ISC) parameter is presently available as part of the CNAV navigation message [7]. However, no public evidence has been provided so far, that the biases between the L2C signal components are indeed small enough to be neglected in practice. For precision applications in geodesy and surveying, distinct DCBs for the various L2C components should therefore be derived from actual observations.