Questionable GSM Tower Locations

Archives Location
#1

A long while ago I purchased an app called Net Cell Info. Over the years of using the app I noticed that a very high percentage of the cell tower locations were reported incorrectly.
Many locations say that the cell tower was in a certain location where there was residential housing or an open field. I contacted the developers of the app and they have pointed me to Mozilla Location Services. They said that were using your services for their app.

Is there a way for me to qualify the details of the cell tower locations?

Thanks
Lawrence

#2

Hi Lawrence, thank you for reaching out.

Mozilla Location Services (MLS) is designed to estimate where a device is, based on what WiFi and Cell Towers it can see. Our model for cell towers is centered on where devices see that cell tower, which rarely includes the cell tower itself.

Our observations documentation explains some of these details, including a diagram of a MLS cell tower model located a far distance from the physical cell tower, because the observers are mostly along a rural road:

image

If you are interested in the physical location of cell tower, the OpenCellID.org website tracks these as an open database:

https://opencellid.org

If you are primarily interested in the physical locations, you may want to look for an app that uses that data source.

#3

Dear John,

Thanks for your response, but I am very curious.

If devices estimate where towers are, then how can any field test and the resulting statistics be reliable. I used to work for a company called Comopt, they did mobile frequency planning so I know a fair bit about all the stats and how they are gathered.

Is it such a huge issue to get the exact locations from the telco providers so that the locations can be set and accurate.

I will look into opencellid

Thanks

Lawrence

blocked.gif

#4

I may have been unclear. MLS tracks where you observe a cell tower, and does not try to determine where the tower is. Maybe an analogy would help?

In my hometown, there is a skyscraper that has one logo on the north side, and a second logo on the south side. If you look up and see logo 1, then you are north of downtown, and if you see logo 2, you are south of downtown. If you are inside the skyscraper, you see neither logo, and have to use other landmarks to tell where you are.

In a similar way, a cell tower has several radios, which broadcast over narrow wedges, and are tuned to provide coverage to certain areas. If you can see a wedge, it means you are in the targeted coverage area. MLS tries to estimate this coverage area, and does not concern itself with the physical location of the tower.

You could imagine a location system that estimate locations based on triangulation from known tower locations. There may be advantages to this system, especially if you can get accurate tower position from regulatory or industry sources. This is not what MLS does, however.

#5

John,

As I said I worked for a company that did mobile frequency analysis and frequency planning. So I do know how towers transmit. As you say, from the tower, there is a triangular broadcast area, at the top of the triangle, there is an arc, and not a definitive line. So what you are saying is that when the phone gets into the broadcast area, it reads the broadcast power delivered by the tower, and it does a guestimation of the tower location (as far as I can gather the tower location is of no significance), what is important is the power received from the tower.

This would make more sense to me.

Lawrence

blocked.gif

#6

Good, you probably have a deeper intuition about how this works than the typical MLS user.

The MLS model of a cell tower is more like a measurement of the circle / ellipse of light made by a flashlight. The MLS cell tower is “located” at the center of that circle, and the radius is how wide that ellipse is. The position of the source (the physical cell tower) is not tracked by the system at all, and does not contribute to the location estimate.

The broadcast power is used to weight a cell signal’s contribution to the location estimate, but not to estimate distance from the tower. MLS does not have a model of power versus distance from source.

#7

John

Thanks for the clarity.

Much appreciated.

Lawrence

blocked.gif