What does GLONASS add?
In the area of vehicle testing most OxTS users work in open proving grounds with few or no trees, no building, nothing to block the sky. In these conditions GPS work perfectly all day long, every day. The number of occasions when GPS will drop from its most precise mode (RTK Integer) to a worse positioning mode is very small.
For ADAS testing in real environments, engineers have to venture out to real roads. This is territory that our surveying customers know well. There are bridges, trees, high-rise buildings, urban canyons. Accurate positioning in these conditions is not as easy. It is difficult to give any specifications for these conditions because:
1) The GPS constellation changes throughout the day, so the same building blocks different satellites at different times.
2) As you drive, different parts of the sky are blocked and so the performance changes along the route.
Both the route and the time of day affect the performance that can be achieved.
Background to GLONASS
September 2010 Russia launched another GLONASS satellite, which will take the total number of operational GLONASS satellites to 21 when it goes live. Having 21 satellites in orbit makes GLONASS a viable satellite navigation system in its own right for most of the globe. When combined with GPS however, GLONASS extends the operating conditions so that precise measurements can be made even when there is partial sky coverage. It also reduces the velocity noise in difficult conditions by providing more satellites that can be used for velocity calculations.
The GLONASS constellation is very new. The oldest satellite was launched in 2004 but all the others started operation in 2007 or later. It is only very recently that GLONASS became a viable service and started making real improvements to GPS.
For 9th September 2010 the coverage for the GLONASS system is shown in the diagram.
White areas have coverage between 99% and 100% of the time. Only a few areas have coverage less than 92% of the time and no areas have coverage less than 80% of the time. When the 21st satellite becomes operational then further improvements will be made.
Most of the time there will be at least 5 GLONASS satellites in the sky, at useful positions over most of the world. When combining GPS and GLONASS one additional satellite is needed to synchronise the times between the GPS and the GLONASS clocks; so you cannot just add all of the GPS satellites to the GLONASS ones. With 5 or more GLONASS satellites in view at least 4 satellites are available to aid GPS.
Aiding RTK GNSS
Why does having more satellites help RTK so much?
When your receiver is in its RTK positioning mode then each satellite doesn’t contribute a great deal. In the open-sky case there is not much improvement when you have 12 satellites compared to 6.
However, it is different when your GPS receiver comes out from under a bridge. Now you need 4 (or 5 using GLONASS) satellites in order to setup and maintain your ambiguities. You need an extra satellite to start solving your ambiguities. For the RTK ambiguity search algorithm, the first 4 or 5 satellites don’t help. Having 12 satellites is twice as good as having 8.
With GPS users typically have between 8 and 12 satellites. However, when you add GLONASS there are typically between 14 and 20 satellites in view. Even though one satellite in the GPS+GLONASS constellation is needed just to make the two systems compatible, the number of satellites is still much larger than GPS alone.
These extra satellites help by:
1) Allowing quicker RTK resolution after bridges and other obstructions.
2) Allowing the RTK to be maintained when much more of the sky is blocked.
Showing position improvements using GLONASS
As we have discussed, in open sky environments GLONASS isn’t likely to help since GPS already gives full accuracy almost 100% of the time. The tests here are for relatively open sky conditions with one bridge and trees starting to encroach over the road in places. Using this we can assess the amount of time it takes to relock and the percentage of time that the receivers remain locked.
8 laps of the course were driven, making 16 interruptions from the bridge. The duration of the test was about 25 minutes. During the time of the test the sky-plot for the GPS satellites is shown in this diagram.
The test ran from about 14:15 until 14:40 when there were mainly 8 GPS satellites in view. Although there were 8 satellites in view for most of the test G04 and G09 are always relatively low in the sky and in directions where there are usually trees. The graph below compared the number of satellites tracked by the RT3002 (GPS only) and the RT3002G (GPS+GLONASS) receivers.
The red line shows the number of GPS+GLONASS satellites tracked by the RT3002G and the blue line shows the number of GPS satellites tracked by the RT3002. The cursor lines have been placed at 8 and 4 satellites, which is the band where the GPS-only receiver operates during this test. Had the test been conducted in open sky then 8 satellites would have been in view all the time.
The number of satellites on the GPS+GLONASS receiver varied from 14 to 9 during the test (apart from the bridge). This is significantly more than the GPS-only receiver, especially for RTK applications.
The relock times of the RT2004 is affected by the distance from the base-station far more than the RT3002 and the RT3002G. To help the RT2004 the base-station was placed on the bridge, which is the best place available to help it relock.
This graph shows a histogram of the relock time after the bridge for each of the receivers.
The RT3002G, GPS+GLONASS with both L1 and L2 frequencies is clearly the best. It only took more than 10 seconds to relock its RTK solution once. Second best was the RT3002, with the RT2004 trailing behind in third place. These are the results that are expected.
We can also look at the amount of time each receiver spent in the RTK Integer mode (with locked ambiguities and maximum accuracy). This result is very environment dependent and results between different places cannot be compared.
Surprisingly the RT2004 performs better than the RT3002 here. This is because there were several places along the route where the RT3002 dropped out of its RTK Integer mode because of the trees. In contrast, once the RT2004 had locked its ambiguities it rarely dropped them apart from at the bridge. This is not surprising since, in general, the L2 signal is only used to resolve the ambiguities; it does not help maintain the ambiguities once they have been found. Also, the L2 signal is always blocked at the same time as the L1 signal (because they are in the same direction and because the L1 signal is needed to lock on to the L2 signal). By contrast the L1-only RT2004 has more satellites and they are more spread across the sky; since only the L1 signal is required to maintain RTK lock, the RT2004 performs better once it has relocked.
Each receiver can also make an error when it tries to resolve its ambiguities. For static systems it is very easy to spot errors; the antenna position is known so you have a reference. In this test we do not have a trusted reference that can be used to figure out whether a receiver has the correct result or not. By comparing the receivers we concluded that the RT3002G probably made one error whereas the other two receivers made two or three. These errors were quickly resolved but did affect the inertial navigation system results.
Interestingly the RT2004 appeared to perform better than the RT3002 in velocity measurements too. The additional satellites that the RT2004 could measure helped keep the velocity noise down. This, in turn, reduces the drift rate of the inertial navigation system when GNSS signals are not available and reduces the noise in position when the GNSS signals are available. Since no absolute reference system was available we cannot be sure how large the impact of the additional signals is.
Given the current GPS and GLONASS constellations there is now a clear advantage when using GPS and GLONASS together for high precision positioning in mobile applications where there are obstructions to the sky. The GPS and GLONASS receivers are able to maintain RTK Integer at times when GPS-only receivers cannot. The dual-frequency GPS and GLONASS receivers are able to resolve ambiguities faster. GLONASS is now a viable solution throughout the world and the number of occasions where it is not useful is now very limited.
Customers who are working in non-ideal conditions should consider spending the additional money upgrading their RT products so that they include GLONASS if they want to achieve the highest levels of accuracy.
The Google images in this document have been created from a licensed version of Google Earth Pro. Copyright of the images remains with Google.