China Begins Using New Global Positioning Satellites 168
cswilly writes with the news that China's satellite navigation system, called Beidou, has been successfully activated. "With ten satellites now, 16 in 2012, and 35 in 2020, China is making damn sure they are independent of the U.S. military's lock on GPS. According to the article, 'Beidou, or 'Big Dipper,' would cover most parts of the Asia Pacific by next year and then the world by 2020.'" The BBC also has slightly more detailed coverage.
Re:Better coverage through multiple systems (Score:5, Informative)
It doesn't. The constellation's orbital pattern is uniform across the entire surface of the Earth.
Your John Deere dealer is a little shaky on how GPS operates. The birds are in 12 hour sidereal orbits, which means the pattern (as seen from an fixed location on Earth) repeats every 11 hours and 56 minutes.... Which means (if such an effect as he describes existed) it would steadily and regularly drift earlier through the day. Thus not only would the effect be seen 'about 6PM' every two weeks or so, but it would also be visible at varying times through the day for a week roughly every other week. (This also implies the gaps drift across the Earth's surface in a regular pattern, and would be visible in places other than the northern latitudes.) In addition, he may not realize that GPS accuracy *normally* varies somewhat over spans of a few hours as the geometry of the visible portion of the constellation varies. So what he's seeing is something else, amplified by observer bias.
Yes, they do. But the system is designed and operated such that having a bird offline for maintenance degrades total system performance by only a very small amount.
The problem isn't the GPS system. The problem is John Deere is trying to use the system at an accuracy (100% availability at 1m) greater than the specified [civilian] performance levels (95% availability at 7m).
Re:Is it possible to combine systems in a receiver (Score:5, Informative)
Yes. This has been done for many years in survey equipment. a typical combination of Navstar (U.S. GPS)/GLONASS increases the number of satellites in view and therefore the accuracy. The biggest problem with combinging Navstar and GLONASS is that Navstar is CDMA (code division multiple access) while GLONASS is FDMA (frequency division multiple access). The former technique makes each satellite use a different "language" sort to say, while the later one uses different frequencies. The result is that a dual receiver needs two independent receivers, making them more expensive. New GLONASS satellites will start using CDMA signals in addition to the FDMA, so that legacy receivers work, and some time in the future new CDMA receivers can use both Navstar and GLONASS with a single type of tuner. Galileo was from the ground up designed to use CDMA and as a result, it is much easier to design a Navstar/Galileo dual receiver. As a matter of fact, many survey devices designed for Navstar can be upgraded via a firmware update to use Galileo as well. You can't upgrade to use GLONASS with a simple firmware update, you also need another tuner.
Regarding accuracy, the thing is that you can't go much less than 5m by just adding more satellites. This is because this error is part of ionosphere delays, and more satellites can't correct this error. It is like trying to do a measurement by averaging 1000 readings, but all done with a bad ruler. At some point, you need to figure out how good your ruler is. And the problem is that this changes dynamically so standard Kalman filter techiques also stop being effective for better than 5m accuracy. There are two approaches for this: the first one is dual frequency, and this is in part how Galileo achieves better accuracy. The idea here is to exploit the dispersion property of the ionosphere. It works like this: different frequencies have different delays, so you send the same signal using different frequencies, measure the delay different, and solve for the ionosphere error. This is what survey-grade equipment do, but they do this by tracking the encrypted military P(Y) code, which is encrypted. The result is a dual frequency solution but full of hacks that make it unstable. This means, as soon as the signal is interrupted for a short time, you need to re-sync.
The other approach for sub meter accuracy come from differential GPS. This technique uses to close receivers, one with a fixed known location. By measuring the error on the known location, you can apply corrections to the moving rover. But for this you need a link between the two (radio, UMTS, GSM, etc) or some post-processing. In addition, you need receivers capable of recording RAW data and then doing some complex math.
The cream of the desert comes from using carrier-phase measurements. With this technique you can go up to cm accuracy. This requires tracking the actual carrier wave, and a very precise model of the earth or post-processing software. The accuracy comes at a price: very very unstable. You need clear blue sky and uninterrupted signals. Plus about 20 seconds to lock the signal, even after small interruptions.
So to answer your question: more satellites guarantee better consistency and readings, particularly in cities and urban landscape. But you can't go below 5m unless you enter differential GPS or dual frequency measurements.