[Top] [All Lists]

Re: [Ecrit] Signed LoST mappings

Subject: Re: [Ecrit] Signed LoST mappings
From: Henning Schulzrinne
Date: Mon, 30 Jul 2007 17:59:50 +0200

Richard Barnes wrote:

Yes, the idea would be to have a cut-off time specified in the
phonebcp, after which the call would be considered unverified.

You seem to be assuming some kind of call stateful model.

Any call status authentication model that uses any message other than the INVITE will likely require this sort of state. Otherwise, the verifying proxy has no way to know if the call never authenticates until after the call has finished.

Note that this level of state (e.g., keeping track of all the messages in an INVITE transaction) is useful for things like accounting and billing as well, so VSPs may be doing it anyway.

This is not the right forum for this discussions, but if a VSP relies on such capabilities, they are likely to be in (financial) trouble. One of the core assumptions for ECRIT is that we don't mess with SIP call flows or create emergency-specific call flows.

SIP entities other than the end points or a B2BUA cannot terminate calls.

I didn't mean "cut-off time" in the sense of "time when the call is cut off," but rather in the sense of "time when a flag goes up that this call is not an emergency call." A VSP could use this signal to take whatever action it deems appropriate -- initiating billing, filing wire fraud charges, or (if it has a B2BUA in the path) terminating the call.

See above.


Ecrit mailing list

<Prev in Thread] Current Thread [Next in Thread>