AW: [Geopriv] Re: Geolocation Policy Draft Update

From: Tschofenig, Hannes (NSN - DE/Munich) ^lt;hannes.tschofenig@nsn.com>
Date: Thu Oct 04 2007 - 10:32:07 EDT

Hi Martin,

> I was going to disagree with the change, because I always
> thought that the condition was the best part of the draft. I
> changed my mind. The condition isn't ready. Lisa's points
> on usability uncovered deeper concerns about determinism.

That's the impression I got as well.

I am just curious why none of these concerns have surfaced by other organizations working on similar concepts.
Carl Reed provided us pointer to documents. Maybe we should get in touch with these guys and figure out whether they have spend some thoughts on these issues or to check their implementation experience.

>
> One point I'd like to clear up was over the maintainability
> of the condition. I'd like to think that for users who are
> managing geospatial data, assumed access to a map is reasonable.
>
> That aside, the polygon shape is somewhat restrictive and
> quite clumsy. Speaking from experience, user interface for
> polygons is a pain. A circle would be better. In the spirit
> of extensibility, allowing any shape but making one
> particular shape mandatory would be better. A circle is
> easier to manage and understand - 100m from this point. It
> also avoids some of the determinism concerns.

OK.

>
> This leads straight to the heart issue. The draft makes a
> central assumption about the nature of location data. There
> is a clear assumption that the location of the Target is a
> point, with no uncertainty.

True. Please also note that the work on the document started at a time where the PIDF-LO profile document was not there.

I believe that the location shape aspect is more an issue for the condition than for the transformation.

>
> I'd argue that the assumption is bad. In the location
> business, you need to be comfortable with a degree of
> uncertainty. Virtually all results are some sort of
> probability distribution, usually with uncertainty that is
> significant enough to cause trouble. 50 metres 67% of the
> time is pretty vague by many standards - and that's actually
> pretty good in the overall spectrum of results for mobile
> networks. Ignoring uncertainty just leads to bad results.

Please note that PIDF-LO does not have an explicit way to express uncertainty.

>
> That leads to two approaches, both equally valid, but both
> requiring additional work. In the first, you strive for
> determinism and you specify rigid rules for the handling of
> uncertainty. This means documenting algorithms and requiring
> strict adherence. Alternatively, you could start caring less
> about guaranteeing results and build tolerances in wherever
> the data is used.
>
> From this perspective, the group needs to reconsider how such
> things are done. This also applies to loc-filters, where a
> similar function is defined.

Interesting that you mention the Location Filters document. In some sense one could see location-based conditions and the location filters draft as very similar things. Hence, all the comments we received for the Geolocation Policy draft are actually relevant for the Location Filter draft as well. One could also argue that the concepts in the two drafts could be aligned better.

>
> On a related note - the assumption made in the draft means
> that the transformations described can seriously damage a
> location that includes uncertainty. A polygon that is
> transformed, point by point, can actually be totally invalid.
> Furthermore, a receiver can no longer assume that circle and
> similar results are accurate to any degree of confidence,
> since the uncertainty is not appropriately modified.

If the draft does not state explicitly that the input to the transformation step is assumed to be a point then we should do so. When you reduce location information then it does not really matter whether the shape describing the location of the Target is an arc band, a circle or anything else. We could, however, discuss this topic in the draft.

Ciao
Hannes

>
> Cheers,
> Martin
>
> > -----Original Message-----
> > From: Tschofenig, Hannes (NSN - DE/Munich)
> > [mailto:hannes.tschofenig@nsn.com]
> > Sent: Wednesday, 3 October 2007 1:30 AM
> > To: ext Sam Hartman; Hannes Tschofenig
> > Cc: GEOPRIV; Tim Polk; Russ Housley; Chris Newman;
> eric.gray@ericsson.com
> > Subject: AW: [Geopriv] Re: Geolocation Policy Draft Update
> >
> > I am waiting for the group to provide us some feedback on
> the proposed
> > change ...
> >
> > Ciao
> > Hannes
> >
> > > -----Ursprüngliche Nachricht-----
> > > Von: ext Sam Hartman [mailto:hartmans-ietf@mit.edu]
> > > Gesendet: Dienstag, 2. Oktober 2007 17:23
> > > An: Hannes Tschofenig
> > > Cc: GEOPRIV; Tim Polk; Russ Housley; 'Chris Newman';
> > > eric.gray@ericsson.com
> > > Betreff: [Geopriv] Re: Geolocation Policy Draft Update
> > >
> > > Does the WG actually have consensus to remove location-specific
> > > conditions?
> > >
> > >
> > > _______________________________________________
> > > Geopriv mailing list
> > > Geopriv@ietf.org
> > > https://www1.ietf.org/mailman/listinfo/geopriv
> > >
> >
> >
> > _______________________________________________
> > Geopriv mailing list
> > Geopriv@ietf.org
> > https://www1.ietf.org/mailman/listinfo/geopriv
>
> --------------------------------------------------------------
> ----------------------------------
> This message is for the designated recipient only and may
> contain privileged, proprietary, or otherwise private information.
> If you have received it in error, please notify the sender
> immediately and delete the original. Any unauthorized use of
> this email is prohibited.
> --------------------------------------------------------------
> ----------------------------------
> [mf2]
>

_______________________________________________
Geopriv mailing list
Geopriv@ietf.org
https://www1.ietf.org/mailman/listinfo/geopriv
Received on Thu, 4 Oct 2007 16:32:07 +0200

This archive was generated by hypermail 2.1.8 : Thu Oct 04 2007 - 10:32:44 EDT