Re: AW: AW: [Geopriv] draft-thomson-geopriv-3825bis-01.txt

From: Henning Schulzrinne ^lt;>
Date: Mon Nov 26 2007 - 12:02:38 EST

On Nov 26, 2007, at 10:14 AM, Tschofenig, Hannes (NSN - DE/Munich)

> So, what would you do in the context of the RFC 3825 debate?
> Here are the options:
> * Leave RFC 3825 as is
> * Add uncertainty as described in draft-thomson-geopriv-3825bis-01.txt
> * Only describe point locations (for example, by omitting the
> 'resolution' fields).
> * Add a new shape that is able to describe a circle.

I think there are two objectives:

- keep it simple, as anything complicated will likely be either
ignored or mis-configured

- keep it honest, i.e., don't pretend to convey knowledge (such as
confidence) that generally doesn't exist in real-world deployments

- keep it conservative, i.e., it's better to have a slightly larger
coverage area rather than having users be actually located outside the
claimed area.

I don't think circles are significantly better than the spheroid
rectangles that are essentially in 3825 today.

I suspect that from an implementation and operational perspective, the
most useful thing would be a table that tells a sys admin what value
to configure for a typical outdoor AP at various common longitudes and
latitudes. Simply assuming that an unamplified AP has, say, 300 ft
coverage is probably as good as anything we can come up with.
Specialized extremely well-maintained and surveyed systems may be able
to come up with something more precise, but they can figure out the
bit math themselves since they have more than an A+ technician on
staff. It is very unlikely that your local Starbucks or Bryant Park
(free WiFi in NYC) will pay for such a survey.

If I were to redo 3825, I would have suggested a simple circle
diameter, maybe rounded to the nearest power of 2 meters, as that's
easy to understand, but that's water under the bridge.


Geopriv mailing list
Received on Mon, 26 Nov 2007 12:02:38 -0500

This archive was generated by hypermail 2.1.8 : Mon Nov 26 2007 - 12:05:24 EST