Re: Notes from Non-meeting

The current design of the VoIP 911 architecture (to be re-issued
shortly) envisions that the release of any privacy-related information
(IP address, name, etc.) is in-band with the call request, typically
included by the caller (if local user agent policy provides for that).
Thus, the PSAP/emergency call center never asks a third party. This
simplifies the problem greatly. Since the current design designates a
universal emergency identifier ('sos') for VoIP, the user agent can
easily make a policy decision ("only include location information if the
call is destined for 'sos'").

> Authenticating the number is less important because firstly, it may
> not matter: someone called 911 and needs emergency help at this
> location. Secondly, it may not be available. In some jurisdictions,
> an otherwise non-functional mobile phone can be used to call emergency
> services numbers. Thirdly, if caller id breaks, emergency service
> should not. So, if we're going to authenticate the calling number,

The problem is not authenticating the caller, but authenticating the
PSAP.

> and that authentication ever fails, we have two choices. We can cut
> off the call (which I think is a bad idea), or we can accept it
> without authentication. If we accept it with an authentication
> failure, why bother trying to authenticate?
>
> Adam
Received on Wed Mar 20 10:51:20 2002