| General > General Technical Chat |
| Global Positioning System navigating in three dimensions |
| << < (3/3) |
| Berni:
You get a pretty random distribution of GPS satelites across the sky because they have fairly low and fast orbits going in all sorts of directions. But yeah you are definitely not getting any satellites from below since the planet is in the way. Tho GPS receivers will tend to take the signal from the low elevation satellites with a grain of salt, trusting more the high elevation ones. This is because low elevations make the signal travel through a lot more of the atmosphere where varying conditions can cause slight variations in the radio propagation speed, messing the measurements up a lot more. So while yes geometrically these satellites seem more favorable, but physics gets in the way of it. This is the reason why the military L2 GPS signal is on a different frequency band (This makes antennas more difficult). The other frequency is affected by the atmospheric conditions slightly differently, so comparing the L1 and L2 signals can give it an idea of how much the signal was distorted and so allowing it to be corrected for. |
| cdev:
The issue with accuracy in the vertical plane can be reduced substantially in many GPS's simply by turning the GPS on and leaving it on for a while (the longer the better, dont just turn it on and use it when it first aquires its fix, the longer it stays on the better the positional accuracy will likely be.) If you can wait a few hours before using capturing elevation readings you'll get a better reading. Ideally it should be just left siffing there. If you have a kalman filter setting thelling it its being used by a pedestrian you'll get better accuracy too. With a GPS in a cell phone it should have the full accuracy available soon after its turned on. You can get slightly better vertical accuracy by setting an elevation mask of 15 degrees, this tells the GPS to ignore signals from satellites that are near the horizon which often have reflections. Make sure your antenna is sitting on a flat metal ground plane if its a patch type antenna. |
| kjpye:
There seem to be two major problems some people are having here: 1. A lack of knowledge of how GPS really works; and 2. spherical geometry is hard. GPS satellites travel in 6 well-defined orbits at an altitude of a bit over 20,000 km. They are not random, and not particularly fast. The orbital period is 12 hours, so the same satellites will be in about the same positions every 12 hours. GPS uses the time-of-flight of signals to calculate the distance from each satellite and, knowing the position of the satellites (which the satellites tell you) you can calculate your time and position. A satellite which is overhead will give you a good idea of your elevation but a very poor idea of your horizontal position. (Moving sideways will not change your distance from the satellite much.) Similarly, a satellite near the horizon due north of you will give a good idea of your latitude, but not much else, and so on. Even if the satellites were randomly distributed across your sky, most of them would still be closer to the horizon than overhead. This is fairly obvious if you imagine we are the centre of the earth -- in which case only 25% of the sky is closer to overhead then the horizon, but because we are about one quarter of the way out to the satellites, and satellites below the horizon can't be seen, most satellites are much closer to the horizon than overhead, and so the accuracy of the elevation suffers. |
| Berni:
In terms of low elevation i mostly meant the sort of below 15 degree elevation where it really goes on trough the atmosphere for a long time, or in a lot of cases will be obscured anyway by hills, buildings, trees... Yes geometrically the north south position is more favorable, but you still tend to get a a few satellites in a cone above 30-45 degrees that can provide useful elevation info. Especially on a modern receiver that will make use of GLONASS satellites too. So it would make sense that you would still get the usual a handful of meters accuracy. If the goal is to get accurate altitude then normal L1 GPS is the wrong way to do it anyway. You can set up a stationary reference station and use RTK capable receivers to do differential GPS and get millimeter accuracy. With long distances between receivers it gets more into the centimeters similar to military L2 receivers, but still orders of magnitude more accurate. But for cycling? Who cares. Not like you would feel any difference between cycling up a 300m or 290m hill |
| cdev:
The GPS system is like a tape measure, it tells you how long that tape measure is at a precisely defined time, once with each satellite. The GPS also has access to a ephemeris which is orbital predictions for the many SVs (space vehicles, basically GPS and WAAS satellites) WAAS can improve the accuracy of altitude signals, BTW. Also the GPS has access to a "geoid" model which is a model of the oblate spheroid of earth (its shaped like a slightly squashed ball, its not perectly spherical) and also earths tides, and also gravity anomalies figure in there too. Basically, GPS is one of the best examples out there of engineering, using multiple disciplines in physics, electromagnetics (radio), and mathematics, making heavy use of a great many branches of science together. If you have access to a GPS with the ability to be used with it and want to see all this being done, I highly recommend RTKlib a program by Tomoji Takasu, a Japanese professor who single handedly revolutionized the affordability of ultra precise positioning.. His web page is at http://gpspp.sakura.ne.jp/indexe.html Its possible that your garden variety GPS now supports RTKlib out of the box and using RTKlib can have sub centimeter accuracy, with a good antenna. many cheap Ublox receivers do to varying degrees. |
| Navigation |
| Message Index |
| Previous page |