30

If my understanding is correct, a GPS satellite produces a pretty simple signal which is basically composed of its location and its time. Given 4 of these signals, then one can uniquely solve for the X,Y,Z position (and as a by product, time) of the hand-held GPS unit.

Why don't we use use cell towers like we use GPS satellites?

It's much more easy to tell the position of a cell tower than a satellite (they don't move). And they can be given atomic clocks just like GPS satellites are given. Then we would have more redundancy, more availability, and more accuracy in determining location.

Note: I know that the E911 service uses cell tower to triangulate the position of cell phones, but this technology is based upon measuring the signal strength of the tower and is therefore nowhere nearly as accurate as GPS.

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
John Berryman
  • 649
  • 1
  • 6
  • 8
  • 1
    you can only hang so much on a single tower, and the providers wouldn't want to give up existing bandwidth to the communication. Cost – Brad Nesom Jan 06 '11 at 06:52
  • How about the simple fact of coverage, getting global coverage is much easier with satellites than towers. – r_ahlskog Feb 07 '14 at 14:46

7 Answers7

28

Cell Towers do not cover the globe or remote/rural areas, Global Positioning System does - however in dense urban areas cell triangulation can be better than GPS.

http://en.wikipedia.org/wiki/GSM_localization#Network-based

Atomic clocks are expensive to make. A GPS receiver works by measuring the relative time delay of signals from a minimum of three, but usually more GPS satellites, each of which has three or four onboard caesium or rubidium atomic clocks

http://en.wikipedia.org/wiki/Atomic_clock

Mapperz
  • 49,701
  • 9
  • 73
  • 132
  • 8
    +1 Good point about the price. A cheap atomic clock costs $50K (http://www.clocktypes.com/buy_atomic_clocks.html ) and a cell tower costs $75K to $200K (http://www.squidoo.com/build-a-cell-tower ), so three clocks would usually cost more than the tower itself. Given that geolocation is an incidental purpose of cell transmission, such a cost increase would be hard to justify. I also suspect the signal corrections for transmission close to the ground would be more challenging to carry out than satellite corrections. – whuber Jan 06 '11 at 03:36
  • 1
    @Mapperz Interesting point about price that I had overlooked.(...which is a shame. I thought my next pocket watch was going to be a "Super Cool Caesium-133".) However, let's get around this by having the towers constantly reestimating their time based upon their connection to GPS. They wouldn't be nearly as accurate, but they would be accurate enough to provide more information. – John Berryman Jan 06 '11 at 12:58
  • @John Berryman - as Brad Nesom mentions Bandwidth Costs - Cell Towers are reaching their limits with 'Data Transfer' requests - Ever had the 'Network Busy'? re-estimation will cause this occurrence even more. – Mapperz Jan 06 '11 at 15:20
  • 1
    Reminds me of a comment a co-worker once made (paraphrased): we can do anything, given unlimited time and funding. – Jaime Soto Jan 06 '11 at 15:23
  • @John Berryman Brilliant idea! It wouldn't take much processing or bandwidth to obtain and retransmit times (stick them in dead space in packet headers or wherever). But I now find myself wondering how that time information would work. Doesn't it need to be accurate (and consistent with other signals) to about a nanosecond? Wouldn't the digital reprocessing and packaging for transmittal take far more than that and be subject to perhaps microseconds of variation? Maybe a retransmitted clock time wouldn't really improve on the triangulation we can already do today... – whuber Jan 06 '11 at 18:16
  • @whuber - certainly the time accuracy wouldn't be nearly as good as the GPSs' accuracy, and there is not new information created from the towers simply reestimating the time. However, where the towers would provide more information is in the fact that their location can be very accurately determined because they don't move, and the location only has to be determined once. Also, even if you don't have accurate times, if you can at least quantify the uncertainty statistically, then there may be enough information to still get a good fix on location – John Berryman Jan 07 '11 at 04:07
  • @John Understood; thank you for explaining. My concern--which is not based on any engineering knowledge--is if the processing and rebroadcasting of the received times is variable or has a lag, it could lead to substantial errors. After all, a microsecond equals 300 meters: that's a huge locational error. But you're right: this is a matter for testing and statistical quantification. – whuber Jan 07 '11 at 05:12
  • @whuber and you're right... if every tower can help you estimate your position to within 300 meters of truth, then perhaps if you had 1000 towers around you, then you could get a good fix on your location. So there might be very practical issues about the quality of the information you receive. – John Berryman Jan 07 '11 at 16:45
  • Since the cell towers don't move (sure, landslides, earthquakes, rule-making exceptions), I imagine that you could merely find the time difference between the tower's location and the location of a given atomic clock. Then, the clock just has to send its time to the tower, and the tower can adjust that time by the difference and send it out. A single clock could service several hundred towers. Moreover, the networks already send the time to most dumbphones on a semi-regular basis anyways. The data complaint is bogus. – jvangeld Mar 15 '11 at 00:05
  • As said before, the number of cell towers in rural areas is low, because cell operators find that this kind areas are not profitable. In addition, there would be another problem: in flat terrains, one these cell towers can cover an area as large as a 35 kilometers radio circle by just using one omnidirectional antenna (GSM, 850-900 MHz). I guess that in such scenario there would not be a lot of room for accurate geo-location. – dariapra Mar 15 '11 at 19:37
  • Why would you even need atomic clocks on fixed positions ? There is no need. On a sat you do, but putting a transponder on an antenna together with a local database of LAC/CELL codes and their Lat/Lon coordinates on the receiver will give you triangulation capabilities, something any smartphone can do. – Glenn Plas Dec 27 '12 at 01:36
6

I would suggest this is due to cost. The Mobile Telecommunication Market is very cut throat and every service provider, manufacturer and software providers are generally taking losses on handset sales in preference for Plan based monthly payment plans. Because of this the overhead of upgrading towers and handset design to include the required hardware and software is prohibitive until we as consumers demand it.

In saying this localised positioning systems are used, here in Australia the Harbour of Whyalla uses a similar idea using radio receivers at points of known location and there by the changes in the wave length can be used to assign the distance.

Could Work but not likely with current Infrastructure.

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
CDBrown
  • 2,350
  • 16
  • 20
5

There is something in use quite like it though. It's called LORAN (LOng RAnge Navigation). It is a terrestrial radio navigation system using low frequency radio transmitters in multiple deployment (multilateration) to determine the location and speed of the receiver. It fell out of favor though when GPS gained widespread adoption. It still lives on as eLORAN ( enhanced LORAN ) and is supposed to be a backup to GPS. Its future however, is uncertain.

R.K.
  • 17,405
  • 3
  • 59
  • 110
4

While not a direct answer to the question, cell towers are already being used in improving the GPS collected in the field.

Google search for GPS-RTK networks and you'll discover how people are using this technology.

Essentially you have communication between a GPS rover and a GPS base station with the cell phone as the middle man.

With the correct GPS equipment, the result can be real-time, survey-grade, centimeter accuracy.

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
3

Well, they actually are used for geolocation (mostly for 911 calls and law enforcement/intelligence surveillance):

But for the reasons listed in other people's answers, it's really not a replacement for GPS, but more of a hack that the Feds can use when they need to stalk someone they don't like, and can't place a GPS device on their vehicle.

J. Taylor
  • 673
  • 6
  • 19
2

There are many more radio reflecting surfaces and obstacles on ground level. The same signal is going to be reflected from different surfaces and received with different delays. I'd suspect GPS signal isn't designed to handle multi path propagation that well.

jkj
  • 161
  • 3
  • 1
    Yes, but this is an issues with GPS anyway. If you try to use GSP in an urban canyon, you can't get a good fix on location due to multi path propagation. – John Berryman Mar 13 '11 at 18:11
2

Assisted GPS technologies used for E911 do use techniques similar to GPS to determine position. Techniques such as Advanced Forward Link Trilateration and Enhanced Observer Time Difference use the timing of signals from cell towers to calculate more accurate positions than simple signal strength estimates. Accuracy of around 50m to 200m is possible.

Mark Goddard
  • 21
  • 2