Having just gotten my first smartphone a few months ago, I’m now trying to learn the process of determining which tower its connecting to, and its physical location.
It seems the two main worlds are MLS/OpenCelliD/Unwired Labs, here, and CellMapper.net.
To actively participate I see that I need the Tower Collector app to submit to this project, and CellMapper’s app to submit to them.
Am I correct in noting that one big difference between these two projects is that participants in CellMapper can manually adjust the locations of known towers, while here machine algorithms alone determine the towers locations?
In other words, Cellmapper has its red vs. green notation for estimated vs. known location, while here there is only some sort of accuracy range?
Some preliminary comparisons show that CellMapper’s greens are truly accurate, and their reds are in the general ballbark, while this project’s locations seem more like where the tester is reporting from than the actual cell transmitter.
With CellMapper I can see there is a way to correct their inaccurate locations, but what can I do here?
E.g., for an OpenCelliD entry for a building mounted antenna near a freeway, it seems that many data samples have come from on the freeway, but maybe none from the roads around the building. What could I do to nudge the location away from the freeway back to the building itself?
I’m not going to root my phone, nor so I have engineering screens that give me super detailled info from the modem. So I’m limited to what the Android API can provide.
The guides for the collector type apps advise driving a circle around the cell site. For tower sites along a freeway, this means figuring out the surface streets on both the side with the tower, and across the freeway on the other side. This takes some planning, and being careful to only have the collector app running at the appropriate times.
What I’m wondering is if the readings while on the “behind” side of the tower will be used to compute the actual location of the tower.
In other words, will all the effort to collect readings all around the tower help make the reported location more accurate?
And, if fewer people do this, than those who just drive along the freeway, does the algorithm have any way of compensating?
Therein lies the problem with both projects. Data from OpenCellID will always be highly approximated due to the lack of ability to edit, and the majority of readings coming from freeways. And there doesn’t appear to be the desire from the CellMapper developers to share their data with OpenCellID. The fact that pretty much only inaccurate OpenCellID data is used by most apps makes this problem even worse.