Adding GLONASS to GPS gives a total of about 50 satellites, for a significant improvement in navigation availability, reliability, robustness, and convergence time through a new multi-GNSS precise point positioning (PPP) service. System performance and field results demonstrate that there is no need to await future constellations — better performance is available now.
By Tor Melgard, Erik Vigen, Ole Ørpen, Fugro Seastar AS, and Jon Helge Ulstein, Bourbon Offshore Norway AS
Precise point positioning (PPP) stands out as an optimal approach for providing global augmentation services using current and coming GNSS constellations. PPP requires fewer reference stations globally than classic differential approaches, one set of precise orbit and clock data is valid for all users everywhere, and the solution is largely unaffected by individual reference-station failures. There are always many reference stations observing the same satellite because the precise orbits and clocks are calculated from a global network of reference stations. As a result, PPP gives a highly redundant and robust position solution.
The results presented here represent a significant step forward in PPP GNSS research and development. Using GLONASS improves the availability and reliability of the solution. The G2 system’s horizontal positioning accuracy is at the decimeter level. These results derive from increasing the number of satellites in the constellation by 60 percent, from about 30 to 50 satellites. The outcome of the development of the G2 real-time combined GPS and GLONASS PPP service represents a next-generation GNSS augmentation. Further, the later GLONASS-M satellites have improved performance and lifetime over previous GLONASS satellites, so that results will continue to improve as that constellation is replenished.
G2 development has benefited from the close cooperation between Fugro and the European Space Operation Centre (ESOC), an establishment of the European Space Agency (ESA). ESOC has contributed its long experience and expertise on precise orbit and clock processing techniques, while the strength of Fugro is real-time positioning and navigation services.
Based on this work, Fugro has introduced the first real-time GPS and GLONASS precise orbit and clock service. The service utilizes Fugro’s own network of dual-system GNSS reference stations to calculate precise orbits and clocks on a satellite-by-satellite basis for all 50 satellites of the two global navigation satellite systems. The system comprises about 40 dual-frequency GPS and GLONASS reference stations distributed around the world as shown in Figure 1.
Raw GNSS measurement data for all satellites are transmitted to processing centers for calculation of the precise orbit and clock of each GPS and GLONASS satellite (Figure 3). The precise data generated is then broadcast to users via geostationary communications satellites with nearly global coverage, as shown in Figure 2.
Inside the end-user equipment a dual-frequency carrier-phase-based PPP solution gives horizontal positioning accuracy at the decimeter level. The PPP calculation module is provided by Fugro and is embedded in multiple GNSS receiver manufacturers’ products as well as Fugro’s own product line.
Like any GNSS technique, PPP is affected by satellite line-of-sight obstructions. Even the most precise orbit and clock data is useless if the user cannot track particular satellites. When satellite visibility is partially obstructed, a best possible service can be ensured by using the full range of satellites from both the GPS and GLONASS systems. This can occur during a survey of a dense urban environment, and for urban positioning in general. It can occur under heavy tree cover, when a cruise ship is in a high-sided fjord, when an offshore vessel is close to an oil rig or platform, or during ionospheric disturbances.
The trend clearly lies towards increasing availability of GNSS satellites on orbit; many studies predict the future benefits of combining the constellations of GPS and Galileo. There is no need, however, to wait for future constellations to reap the immediate benefits of access to additional GNSS satellites. The current GLONASS constellation may not have all the features of future GNSS systems, but it is available here and now. Recently, the Russian government has proven its commitment to enhancing the GLONASS constellation. Many receiver manufacturers have also acknowledged this fact and now provide combined GPS and GLONASS receivers.
G2 Accuracy and Statistics
In Figure 4, time-series plots show the 3D accuracy of GPS and GLONASS G2 real-time orbits on August 14, 2009. In the comparison, final orbit data from the International GNSS Service (IGS) is used as reference. PPP positioning is mainly affected by the radial orbit error, which is significantly less than the total 3D error shown here. The 95 percent 3D accuracy for GLONASS (22 centimeters) is more than double that for GPS (10 centimeters). The graph demonstrates how this difference in this case is mainly caused by a few GLONASS satellites being less accurate. Actually, several GLONASS satellites have orbit accuracy very close to the level of GPS for real-time G2 data.
Figure 5 shows the clock accuracy of the G2 real-time clocks compared to final IGS clocks. A constant bias has been removed to account for the differences in system reference time. Smaller individual clock biases for each satellite can still be observed. Small biases do not affect the final accuracy of the PPP solution, and achievable position accuracy with these clocks are significantly better than the 21-centimeter 95 percent number for GPS may indicate.
The lower time series in Figure 5 shows the estimated GLONASS clock accuracy. Currently there is no comparable IGS product with precise GLONASS clocks. A post-processing of all available IGS plus Fugro GNSS stations has been made to establish a reference for the comparison. As shown, the GLONASS clocks are more variable, but still they are stable enough to allow for precise navigation.
Real-Time Positioning Results
Real-time position performance is continuously observed at the G2 operation and monitoring center in Oslo, Norway. The graph in Figure 6 shows typical G2 positioning results with the calculation engine running in dynamic mode at a fixed location for a 24-hour period. The blue lines in the north and east time series are at 20 centimeters and the scale is 61 meter. In the height graph the blue lines indicate the 30-centimeter level. The antenna is in a location with clear view of the sky, and in
dependently calculated reference coordinates are used as reference. 1-sigma accuracy statistics on August 14 are 3, 4, and 8 centimeters in easting, northing and height respectively.
Figure 7 shows GLONASS-only real-time positioning with clear view of the sky for the same day as in Figure 6 and the same antenna location. The blue line indicates the 50-centimeter level and the scale is 62 meters. For long periods, the GLONASS-only solution works quite nicely. There are, however, shorter periods with fewer than four satellites being tracked, causing the position output to stop, followed by a period of re-convergence.
Figure 8 displays results from May 11, 2009, when there were slightly more satellites available and just enough to have the GLONASS-only solution running for 24 hours without resets. 1-sigma accuracy statistics for this day are 11, 9, and 16 centimeters in easting, northing, and height respectively. Considering the average number of satellites of 6.14 and periods with high DOP values, this is very promising. In early 2010, 20 GLONASS satellites should be available, and by 2011, 24 are expected. In 2010, a performance similar to or better than that of May 11 should generally be expected with the new satellites. By 2011, even better performance is believed to become the norm of GLONASS-only real-time PPP navigation.
Even in some clear-view-of-sky situations, the addition of GLONASS may improve the navigation compared to GPS-only solutions. Figure 9 presents an example of such situations. Here the GPS-only solution suffers some multipath-like effects showing up, especially in the east component. Figure 10 shows the combined GPS+GLONASS solution for the same dataset. The distortion in position is practically eliminated. This is an example where adding GLONASS also improves redundancy and accuracy for navigation with clear view of the sky.
The next test further analyzes the same dataset as in Figures 9 and 10 by simulating a virtual wall to the south, blocking all satellites below 40 degrees elevation. Figure 11 illustrates this virtual wall blocking both GPS and GLONASS satellites.
With such data blockage, the GPS-only solution fails for more than 20 minutes, as seen in Figure 12, simply because the number of satellites goes below four. Then a period with slow convergence follows because of few satellites and high DOP.
Again, adding GLONASS greatly improves the performance, as shown in Figure 13. Now a sufficient number of satellites are tracked all the time, and there is a continuous solution with the combined GPS+GLONASS throughout the time window when the GPS-only solution failed.
Even with more than 30 satellites in the GPS constellation, there are situations when the satellite geometry gets poor. This occurred in northwest Europe on February 2, 2010. One of the GPS satellites (PRN17) was not available due to maintenance, and even with five to six usable GPS satellites left, the horizontal dilution of precision (HDOP) was in the range of 7–11 for about 12 minutes (10-degree elevation mask), as shown in figure 14. Such high HDOP values lie above what most user installations are configured to accept, and Fugro received feedback from clients at sea losing positioning. The G2 solution was not affected by the poor GPS geometry and kept the HDOP below 2 during this period, as shown in Figure 15.
As will be shown in the following analysis, adding GLONASS not only improves availability and robustness of the solution, it greatly improves convergence time. Real-time high-accuracy PPP solutions use carrier-phase measurements to achieve high-accuracy positioning. To do so, the carrier-phase ambiguities must be determined. This process takes a certain time depending on the observed satellite geometry and is commonly referred to as cold-start convergence time.
Figure 16 presents a theoretical study of the expected convergence time for a GPS-only compared to a combined GPS+GLONASS solution. The lower graph shows how the expected convergence time varies significantly for a GPS-only solution throughout the day, with a peak of 75 minutes. The combined solution shows much more consistent performance, with expected 50–60 percent average improvement over GPS-only.
We compare this theoretical study to results using G2 data produced in real time in Figure 17. A cold start is performed every 5 minutes throughout the day, for six consecutive days, giving a total of 1,728 convergence tests. The convergence criterion is the time when the 3D position arrives within 40 centimeters of the reference position and remains there for a minimum of 10 minutes. The average convergence time improvement achieved in Figure 17 is 39 percent, with some variations from day to day. On the better days, the average improvement is almost 50 percent, and close to the expected performance based on the theoretical study. On other days, there is room for further improvement. Mainly two factors are expected to contribute: more and newer GLONASS satellites, and further improvements of the G2 precise GPS and GLONASS orbit and clock product.
Dynamic Environment Results
Since late 2008, the G2 system has been installed on the vessel Bourbon Topaz, making frequent trips into the North Sea and back into port in Norway (see BOX).
All positioning data from both the G2 system and the GPS-only reference systems are logged in real time on the vessel. Figure 18 gives an example plot of the relative height estimated by the G2 GPS-GLONASS solution. In the beginning of the plot, the vessel is out at sea, clearly seen as a noise in the graph that actually is the vessel’s movement in the waves. Then the vessel comes into port and the slower tidal variations are observed for the next 12 hours until the vessel again goes back out to sea.
On June 22, 2009, an incident was recorded where the combined GPS-GLONASS G2 solution improves performance. As seen in Figure 19, there is a period starting at 10:00 UTC where the GPS-only reference systems suffer from poorer DOP values, and this is reflected both in horizontal and vertical components of the calculated position. This particular plot shows how the height drifts off by roughly 1 meter while the G2 combined solution remains unaffected for the entire period. Generally, the G2 solution also shows a smoother height than the reference system even when such problems as shown here are not present.
Test of G2 onboard Bourbon Topaz
The Bourbon Topaz is a modern supply vessel equipped with the latest dynamic positioning (DP) systems, operating in the North Sea. The North Sea can be a harsh environment in which to operate, and we rely on good tools for maneuvering our vessels.
Early on, we recognized the need for stable, reliable reference systems, and our fleet is equipped with Kongsberg Seatex DPS700 system as standard. When we were asked to test the G2 onboard the Bourbon Topaz, we saw this as an opportunity to follow the development in the industry of such services. The DPS232 receiver was set up in connection with the vessel’s DPS700 system, and all information was logged and sent to Fugro Seastar.
We often experience that the vessel has to operate close to offshore installations, which could block good reception of signals. In these cases, the G2 offers a much better and more reliable signal reception. Our experience of the quality of the G2 system is overall positive.
G2 and the other Fugro services can be received from a variety of different user equipment; both Fugro-branded or manufactured equipment and third-party equipment. In most cases the L-band receiver decoding the data from the geostationary satellites, including Fugro subscription software and position calculation module, is integrated into the same box as the GNSS receiver. Both the GNSS and geostationary satellite signals can be tracked with a single antenna.
Test results confirm decimeter-level position accuracy in real-time navigation with G2, the first real-time combined GPS and GLONASS PPP service. Several examples show how G2 improves availability, robustness, and convergence time compared to GPS-only positioning.
More is better. There is no need to wait for future constellations like Galileo to reap the benefits of access to additional GNSS satellites now.
Equipment supporting Fugro services includes receivers from Kongsberg Seatex for marine applications (Seastar), and NovAtel, Trimble, Topcon, Sokkia, Hemisphere GPS, Novariant, and Raven for land applications (Omnistar).
Tor Melgard is R&D manager at Fugro Seastar in Oslo, Norway. He holds an M.Sc. in electrical engineering from the Norwegian Institute of Technology and wrote his thesis at the Department of Geomatics Engineering, University of Calgary.
Erik Vigen is a senior developer at Fugro Seastar. He received his M.Sc. in Geodesy from the Norwegian Institute of Technology.
Ole Ørpen is senior scientist at Fugro Seastar. He received his M.Sc. from the Norwegian Institute of Technology in electrical engineering.
Jon Helge Ulstein is IT superintendent at Bourbon Offshore Norway AS, a subsidiary of the Bourbon Group, Marseilles, France.