Monday, September 3, 2007

IMS, 3GPP and IETF: A standardization complexity

By Tsahi Levent-Levi

How do we get those specifications for IMS? In a complex way.

It started off as a set if requirements for a Next Generation Network (NGN). The 3GPP wanted an all-IP network for its mobile infrastructure, calling it IMS (IP Multimedia Subsystem). As there’s no need to reinvent the wheel, the 3GPP decided to select an existing standard to do the work, and SIP was there – all young and fresh. But SIP is an RFC. It is handled and standardized by the IETF. This need not be changed.

So what does an organization like the 3GPP does at this point in time? Use the IETF as a subcontractor.

Have you ever worked with a subcontractor? I have never heard of anyone who liked the experience… you provide requirements for a rocket to space, and you get a fire cracker. You want a match, and you get a rocket instead. Time is not time, effort estimations are far from true (sounds like regular development, but it is always harder with a subcontractor).

So we have the 3GPP providing the requirements, while the development of new RFCs (=standards for IMS) done by the IETF, including modifications to RFCs when needed.

The result?

  • We have a whole lot of RFCs coming from the IETF. Some colliding each other, others solving the same problems, but a bit differently.
  • We have a bunch of 3GPP specifications, which point to RFCs (and a lot of drafts!) that are used by the 3GPP’s IMS network – in a way, a selection of the RFCs that are needed.
  • But then, it is not always understood which features from the IETF, or the 3GPP you really need to build an application. And as usual, I haven’t covered GSMA, GCF, OMTP and other organizations.

We at the IMTC IMS AG are actually facing these issue each day. We are currently unraveling the set of specifications required for the implementation and interoperability of the Video Sharing service that is gaining momentum.

No comments: