Help needed with GSM Geolocation without GPS

[robin]

> Sounds good.  I wonder what the trade-off is between implementing
> something like this from scratch for GTA04, and trying to integrate an
> existing partial solution such as GeoClue?

hi neil,

as far as I understand geoclue comprises a communication between a provider
and your phone, so that would mean data transfer via the modem. This would
certainly be one solution, but it would be nice if the user could choose 
between:

a) GSM -> Provider -> Location
b) GSM -> offline database -> Location (saves money and battery(?))

I don't know if geoclue can be easily extended to use the cells.txt.gz file 
as an alternative input. That would be nice, because then one could actually
choose between a) and b).
 
> > So I started with step 1 and I am a bit stuck. The numbers for the location
> > that the Mokofaen theme displays somehow do not match any of the cell towers
> > my phone is most likely to be connected to:
> > Most likely I am connected to 
> > mcc 262; mnc 3; lac 40096 and cellid 137380532 [2]
> 
> How do you get those numbers?  They look like UMTS ones (which are
> bigger than GSM).
> 
> > which has nowhere the numbers I get on my home screen:
> > 296/25326
> 
> Those numbers look like GSM.  Are you actually connected by GSM or by
> UMTS?
> 
> (Some background on the GSM/UMTS difference is here:
> http://lists.goldelico.com/pipermail/gta04-owner/2012-September/002923.html)
> 

Thanks for pointing me to the link / I vaguely remembered reading something 
about this on the list. I also found the gsm-cell which corresponds to my 
numbers so all fine.
Regarding the consistency of the databases one apparently has to be a bit
careful. I checked the cells around me with the openbmap map interface, and
there are certainly cells marked as standard GSM which have to be UMTS cells
according to the long numbers they have for their idenfication. But I would
think that this would only matter, if their signal strength rate/decay over
distance is different from the GSM cells. Then one would indeed need to know
what cell type it is to apply the correct function to estimate the distance 
to it. Otherwise the more cells the better it should be.

Could you point me in the right direction on how to extract the cell and
neighbouring cell ids/signal strengths in qtmoko.

br

roibn