Use known wifis to georeference cell towers

[GoogleCodeExporter]

By now, most rail and road tunnels have their own cellular base stations. 
However, since these are areas where we cannot get a GPS signal, we're never 
going to georeference these cells using GPS alone. (I made a few attempts 
walking out of a subway station very slowly, but the device was usually faster 
to hand over to an "outside" cell than to pick up a GPS signal.)

On the other hand, these underground cells are potentially the most important 
ones for geolocalization, being located in areas where no other navigation 
services are available.

So why not use wifis for georeferencing here? Here's how it could work:
1. You have logged the surroundings of a tunnel entrance, subway station or 
similar, using GPS to georeference wifis, and added these wifis to your wifi 
catalog.
2. Now you are underground, being served by the tunnel cell. You have no GPS 
signal, but you are still picking up some of the wifis you have previously 
captured on the surface.
3. Radiobeacon uses the georeferenced wifis to estimate your position and can 
thus store a measurement for the tunnel cell – either using an external 
location provider or by simply looking it up in the wifi catalog.

Georeferencing using wifis is somewhat less precise than with GPS, but since 
cells cover a much larger area, positional accuracy should be less of a issue. 
Also, underground cells cover much larger areas – while on the surface I get 
handed over to a different cell every few hundred meters, I am seeing some 3 
cells on my 8-km commute to work – which means the position inferred from the 
cell is going to be much less precise anyway. Lastly, even a crude position is 
better then none at all.

Currently, when I am sitting on the subway and watching SatStat, I get position 
updates fat most stations, presumably from the wifi provider. The confidence 
interval looks quite decent and accuracy also looks good – far better than a 
location obtained using cellular networks – therefore the system should work.

Original issue reported on code.google.com by mich...@vonglasow.com on 15 Mar 2015 at 3:37

[GoogleCodeExporter]

We might give it a try, but it will require significant changes to the logging 
service, so no time schedule for release yet.. 

Another possible approach: some very long time ago, I experimented with a 
combination of step counter and compass orientation to get measurements indoor 
where no GPS is available. The code was functional, but the results where 
somehow disappointing due to the sensor drift. It may nevertheless produce a 
rough estimation of current position..

Original comment by wish7code on 15 Mar 2015 at 6:45

• Changed state: Accepted

[GoogleCodeExporter]

I've experimented with intertial navigation myself a while back (actually 
SatStat came out of a test app I wrote for that) but didn't get very far. 
Figuring out movement based on sensor data is quite complex and error-prone. 
Especially the compass needs a frequent calibration procedure, and still the 
data is frequently off.

I'd say tapping into wifi scan results and looking up the location in the 
catalog is the easier part, and most likely produces more reliable results...

Original comment by mich...@vonglasow.com on 15 Mar 2015 at 7:49

[GoogleCodeExporter]

> I've experimented with intertial navigation myself a while back (actually 
SatStat came out of a test app I wrote for that) 

Maybe we should join efforts.. Next time I'm in your hometown I'll arrange a 
beer ;-)

> I'd say tapping into wifi scan results and looking up the location in the 
catalog is the easier part

I would totally agree..

Original comment by wish7code on 15 Mar 2015 at 8:13

[GoogleCodeExporter]

Beer sounds good :-) drop me a line the next time you are there

Original comment by mich...@vonglasow.com on 15 Mar 2015 at 8:41

[GoogleCodeExporter]

Quite interesting: I wonder what this mysterious (in this case malfunctioning) 
underground gps repeater are...

http://www.ciras.org.uk/report-library/train-operations/52131-issue-with-opening
-class-377-doors-on-the-thameslink-route/

(found via blog.fefe.de)

Original comment by wish7code on 16 Mar 2015 at 11:15