Multiple GPS units to achieve greater accuracy.

biohzrd · 02-10-07, 01:11 PM

I'm wondering if there currently exists any software that will combine the signals of two or more GPS receivers... say from Bluetooth and/or CF devices in order to give a more accurate position? I know there are the Trimble units that use all kinds of 3rd party solutions to achieve greater accuracy, but they cost thousands of dollars. Although I do not currently have a use for such an idea, I think it would be a pretty sweet application if the consumer/end user already has two Bluetooth or CF/SD GPS adapters lying around the house.

Any thoughts on something like this?

JakeRich · 02-10-07, 09:14 PM

I'm not sure multiple would do better. The DOP (Dilution of Precision) of the fix is governed by the arrangement of the satellites. That means that when the satellites are badly aligned, your fix won't be any more accurate on any one of them because the same skew in signals will happen to all of them. The way the lower end Trimbles work is to do a lot of fixes and average them to get increasing accuracy. The balancing act is to do enough to get a good distribution, but to cut off the input before you get a large excursion that would skew the results. Look at the plot of fixes in GPS101. There are a couple of HUGE excursions that occurred during those hours.

It is an interesting idea, though. Anybody want to try to see if there is a way to do it?

02-10-07, 09:24 PM

I can't see how several GPSs would give a better result than just moving a single one around... which is how they are usually used...

ou2mame · 02-13-07, 06:08 AM

i think you'd be better off investing in an external antenna, rather than another unit. but like what was stated previously, having more receivers doing the same thing, connecting to the same satellites that are out of reach as it is, wouldn't make any difference.

02-13-07, 01:01 PM

I just can't think of a pratical use for this. Geocaching is one hobby with a lot of participants, but even there you far exceed the needed accuracy with the average GPS. If you are trying to find a location with a multi-thousand dollar GPS unit, and the person who gave you the coordinates marked it with a Garmin eTrex, the coordinates are only as accurate as his/her GPS. Also, you have to take into account the effort they made to get an accurate reading.

Maybe this would be helpful in a farming application where the crop placement is tracked over long distances with a close tolerance.

02-13-07, 01:22 PM

The thing that does make a huge difference is WAAS...
Link: http://www.garmin.com/aboutGPS/waas.html

JakeRich · 02-13-07, 03:17 PM

Originally Posted by Noisy Crow

The thing that does make a huge difference is WAAS...
Link: http://www.garmin.com/aboutGPS/waas.html

But even WAAS is also just an average based on the regional conditions. Depending on the size of the region, and your location within it, WAAS may or may not be that much better. On average, it will be, but for any given spot, it might not. The thousands of fixes shown in GPS101 were WITH WAAS, and even then there were large excursions.

DGPS, near an airfield, can be very accurate, but also very limited in coverage. It is designed for the landing aircraft to have a very accurate height fix. WAAS, because it is wide area, has to be the average of multiple DGPS corrections.

axim4hiking · 02-13-07, 05:25 PM

Originally Posted by JakeRich

But even WAAS is also just an average based on the regional conditions. Depending on the size of the region, and your location within it, WAAS may or may not be that much better. On average, it will be, but for any given spot, it might not. The thousands of fixes shown in GPS101 were WITH WAAS, and even then there were large excursions.

DGPS, near an airfield, can be very accurate, but also very limited in coverage. It is designed for the landing aircraft to have a very accurate height fix. WAAS, because it is wide area, has to be the average of multiple DGPS corrections.

WAAS is a regional improvement, but the regions are fairly small. IIRC there are 21 correction locations located throughout the US and the correction signal does not differ greatly between stations. Plus, it is an improvement with essentially no cost. The additional processing required for the DGPS signal is all done on chip, as the broadcast signal specification is very similar to the existing GPS. The chip is already calculating the position based upon multiple signals, so WAAS is just an additional correction factor.

It was a lot more important before they turned off SA. Back then, your position could be off by up to 100 meters. WAAS took that down to 7 meters, which was good enough for 737 autopilots to do multiple experimental hands off landings. That was before LAAS (Local Area Augmentation Signal) at airports was available.

mitchman · 02-13-07, 05:37 PM

I use an Axim X51v and and iBlue GPS for data logging in our race car. The manufacture of the PPC software said that the Axim would handle multiple bluetooth signals without problem.

So I guess the answer is yes, the Axim can handle multiple bluetooth signals. But whether your GPS software will, is another question.

JakeRich · 02-13-07, 06:04 PM

Originally Posted by axim4hiking

WAAS is a regional improvement, but the regions are fairly small. IIRC there are 21 correction locations located throughout the US and the correction signal does not differ greatly between stations. Plus, it is an improvement with essentially no cost. The additional processing required for the DGPS signal is all done on chip, as the broadcast signal specification is very similar to the existing GPS. The chip is already calculating the position based upon multiple signals, so WAAS is just an additional correction factor.

It was a lot more important before they turned off SA. Back then, your position could be off by up to 100 meters. WAAS took that down to 7 meters, which was good enough for 737 autopilots to do multiple experimental hands off landings. That was before LAAS (Local Area Augmentation Signal) at airports was available.

History of GPS.

History of WAAS.

History of LAAS.

GPS came first, then DGPS, WAAS and finally LAAS. All three emulate a satellite signal, so that any WAAS-enabled receiver can do the calculations. In order of accuracy, it's WAAS, then more accurate DGPS and most accurate LAAS, although the difference in DGPS and LAAS is zero.

As you can see from this table, taken from the first article, the largest error is Ionospheric, for which all of these mechanisms correct.

Sources of errors Source Effect
Ionospheric effects ± 5 meter
Ephemeris errors ± 2.5 meter
Satellite clock errors ± 2 meter
Multipath distortion ± 1 meter
Tropospheric effects ± 0.5 meter
Numerical errors ± 1 meter or less

aaron19953 · 02-13-07, 06:52 PM

The short answer is no, if two gps units have an accuracy of 1-3 meters, then they are both going to give you the same result (more or less)

However..

Technically speaking yes, two gps units would give you better accuracy. However, one of the units would have to remain at a fixed location therefore making it impossible for you to go anywhere. The idea is one receiver is locked at a known benchmark like a USGS bench which there are tons of around the US. That receiver would get data from the satellites and transmit "RTK" (real-time kinematic) correction signals via radio or cellular to the other GPS receiver.. There's a ton more involved, but essentially that's what happens.

I work for a surveyor that uses GPS and we pay dearly for this accuracy. Our units (as are most survey grade) are accurate to within the area of the face of a dime.

Hope that helps..

-=<> Aaron <>=-

biohzrd · 02-14-07, 10:30 AM

OK thanks everyone, I guess this topic has been thoroughly brought down.

I guess my initial idea was based on the GPS receivers originating from different manufacturers, and possibly different chipsets. Since they all would technically derive the information from the "same" signals, and apply their own unique algorithms to output the data, this data could be averaged for a more precise coordinate. This might be more apparently useful in hiking applications where tree cover kills the signal for one receiver, while the other(s) still get a fairly decent signal...the application could average the results for a more consistent locational fix, resulting in a better graph for outputting to say, Google maps.

JakeRich · 02-14-07, 10:59 AM

Originally Posted by biohzrd

OK thanks everyone, I guess this topic has been thoroughly brought down.

I guess my initial idea was based on the GPS receivers originating from different manufacturers, and possibly different chipsets. Since they all would technically derive the information from the "same" signals, and apply their own unique algorithms to output the data, this data could be averaged for a more precise coordinate. This might be more apparently useful in hiking applications where tree cover kills the signal for one receiver, while the other(s) still get a fairly decent signal...the application could average the results for a more consistent locational fix, resulting in a better graph for outputting to say, Google maps.

Ok, I can see what you were thinking. The dilemma is that at the frequencies used by the satellites, just a few inches can make a big difference in the signal strength and quality. In fact, at some of the sites on the web I've seen pictures of multiple GPS units all lined up on the dashboard of a car, then the "simultaneous" plots of all the fixes to demonstrate that one or the other is more accurate. That test suffers from the same problem: just a couple of inches--particularly in a noisy or poor environment--will make a huge difference in the signal strength from the same satellite. And since most fix algorithms actually do some damping--to prevent wild excursions--if one of them drifts in a direction because of the signal getting wonky, it'll take longer for that one to "drift" back because of that damping. In your hiking scenario, you'd be better off with ONE really sensitive receiver that works under the foliage cover than to average a good one with a not-so-good one that has more wander because it can't see as many satellites.

Also, when displaying data one of the largely ignored factors in most applications is the specific Datum used as reference. Change the datum reference and the fix will move slightly. Match the datum and the fix and you'll be pretty good, but no guarantee. Also, the earth isn't actually round, or even consistently un-round, so no matter how good the datum, as soon as you move from the reference point there is some assumption being made about the earth shape that will invariably be somewhat incorrect.

Finally, even Google struggles with depicting location. I've used plots multiple times on Google Earth and the picture under the plot moves from session to session. Not much, but there must be some rounding taking place because Google shows the exact same set of lat/lon in different places from session to session.

Bottom line: It's very , very difficult to know exactly where you are unless you want to spend a LOT of money and time.