Hi Philipp,

On Wed, Apr 15, 2020 at 09:18:04AM +0200, Philipp Borgers wrote:

How precise is the timing advance for localization?

As per GSM specs, it's one symbol duration, which is about 500m in terms of distance. OsmoBTS supports a higher-resolution reporting in the A-bis RSL measurement reports (in units of 1/256 of symbol duration). However, the actual achievable real-world precision is likely much less (maybe a quarter-bit), resulting in 125m.

Neels mentioned that "modern" mobile stations equipped with a GPS receiver can report the location to the operator. Could you point me to the right standard?

RRLP (Radio Resource Location Protocol). We did a proof of concept implementation for this more than a decade ago, it was tested at HAR2009 at the time: https://osmocom.org/projects/security/wiki/RRLP https://git.osmocom.org/osmocom-lcs/

RRLP is part of the larger LCS (Location Services) architecture: At https://osmocom.org/projects/osmobsc/wiki/Location_Services it is described how they work in the specs.

The Osmocom implementation bypassed all of the standardized network-internal interfaces, had no SMLC, Lb interface, ... - in the end, only the radio interface to the mobile phone is what's relevant to get positioning.

The related code has not been maintained for 10 years or so, all of the rest of the osmocom infrastructure has evolved a lot ever since - and it only was a proof-of-concept to begin with. So I would assume that considerable time would have to be spent in testing / fixing / stabilizing / integration.