Wifi positioning backends

Tim Lawrenz tim.lawrenz at gmail.com
Wed Aug 1 07:59:27 PDT 2007


Hi,

About me: Tim, from Plazes.com ... nice to meet you all  : )

The lat/lng values we return are only somewhat reliable. Usually, we
have a very good chance that our geocoding service provider returns a
very good location (usually the tolerance lies within a 50 meters or
so for europe and the us) if the user enters a fully qualified
address. But a 'Plaze' at plazes is also valid without any address at
all, only with city set or within a zip code range.

It gets interesting with the option to manually edit the lat/lng
values or use the map to reposition a Plaze: The geocoding could be
wrong or position the plaze in the middle of a city (or even a
country) bacause of a typo in the address, etc. A user then could edit
such a plaze and move the pin on the google map to the exact position
of the router without changing the address itself.

That is why the accuaracy level of maporama is not really useful (and
as far as I can see will not be part of the API results). The number
of fields in the address filled out is a small hint about the quality
of the lat/lng values, but not necessarily correct (a real life
example: we have a user who did not really get the idea, created one
single plaze with a full address and then literally moved that plaze
to Washington, New York, San Francisco, depending on his current
position using the map without changing the address. well, go figure).

Hope I could help, I read this list regularily and hope to be able to
provide help.

Kind regards,
Tim Lawrenz

On 01/08/07, Andrew Turner <ajturner at highearthorbit.com> wrote:
> On 8/1/07, Stephen Wing <stephen at wintoncourt.co.uk> wrote:
> > Andrew Turner wrote:
> > > Regarding some distance estimate, I would think 100m be the top-most
> > > error, b/c most AP's don't typically have a range much beyond that.
> >
> > But does itprecisely locate their APs in the first place - if the
> > location of the AP is only accurate to 500m in the first place, then
> > this may be a bigger limiter than the range of the AP itself.
>
> Good point - Plazes itself doesn't locate the AP's, it just relies on
> users to identify where they are. And Plazes could have gotten the
> info from AP, Cell, or Web - so you may be right in that the error is
> larger.
>
> Of course, the number is fairly irrelevant for this case - which is
> why hopefully any client would pay attention moreso to 'civic
> accuracy'.
>
> >
> > Steve
> >
> > _______________________________________________
> > GeoClue mailing list
> > GeoClue at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/geoclue
> >
>
>
> --
> Andrew Turner
> ajturner at highearthorbit.com      42.2774N x 83.7611W
> http://highearthorbit.com              Ann Arbor, Michigan, USA
> Introduction to Neogeography - http://oreilly.com/catalog/neogeography
> _______________________________________________
> GeoClue mailing list
> GeoClue at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/geoclue
>


-- 
plazes: you know where - http://plazes.com/user/tim


More information about the GeoClue mailing list