End-to-End Session Identification of Multimedia Sessions

One of the challenges we’ve often faced with IP multimedia communications systems is that of session identification. H.323 has the concept of a “Call ID” and SIP has the concept of a “Call ID”, but they are not compatible with each other. As such, it’s impossible to allow a session to be identified end-to-end. This opens up the possibility of getting into call loops, etc.

Aside from avoiding network issues like call loops, though, there is also a desire to be able to identify a session end-to-end (even through interworking functions), track a session as it is transferred, identify sessions that are part of the same multipoint conference, and associate media flows with a signaling session.

In H.323, we introduced a field (called CallLinkage) that made an attempt at tracking calls as they were transferred. However, the rules for how to populate those fields were complex and, in the end, few implemented the procedures. Still, what would work for H.323 did not work for SIP.

Within H.323, we have the ability to associate all devices that are part of the same conference, since all participants use the same conference identifier (CID). Well, so is the theory. In practice, though, most multipoint conferences are really just a logical association of point-to-point sessions. So, the conference identifier in H.323 has not been use consistently. And, as before, what would work for H.323 does not work for SIP.

Management systems, policy enforcement points, and other network elements might also want to associate session signaling with media flows. To do that, somehow those two need to be correlated with each other. There are ways we could allow that to work for SIP or H.323, but there are real-world implementation issues in trying to do that. Further, selecting a mechanism that works for SIP or H.323 would not allow this to work end-to-end.

In recent weeks, there has also been a lot of activity on the SIPREC mailing list in the IETF where folks recognize a need to have some kind of session identifier, and what currently exists for SIP is not workable.

For these (and other) reasons, some colleagues and I put together an initial draft of a new Session Identifier that can be used by H.323, SIP, or even the forthcoming H.325 multimedia communication systems. Feel free to give us feedback!