The Rush to Support Iu-h

There has been a lot of fervor surrounding the 3GPP's activities to develop a new Home Node B (aka femtocell) standard.

I came across the blog post the other day with the clever title "Supporting Iu-h, why the rush?"

The author has a pretty good take on the situation.

RAN Gateway wins again

I have spent some time considering the pros and cons of the new 3GPP HNB femtocell architecture. It’s too early to call it a standard, but for now, the basic building blocks are in place for putting the spec together.

One of the fundamental building blocks of the HNB effort is a ‘RAN Gateway’ style architecture. In this case, the RAN gateway network element sits ‘parallel’ to an RNC or BSC, and connects into the mobile core network via existing, well defined circuit interfaces (Iu-CS) and packet interfaces (Iu-PS).

Operators evaluating femtocells and femto architectures immediately grasped the benefits of using a RAN gateway approach:

• All existing circuit and packet services are supported through the femtocell access network as they are supported through the macro RAN network. This is because the RAN gateway uses the same service interfaces as the RNC.
• A ‘RAN’ gateway is located in the RAN, with a clear demark point into the mobile core network. For operators, they did not need to change the MSC or the service core in any dramatic fashion.

For all the rhetoric and posturing around the benefits of SIP and IMS, operators quickly understood that the most pragmatic approach for femtocell service delivery was through a RAN gateway architecture. In fact, one of the things that the proposals from Huawei, NSN, Alcatel/Lucent and Kineto/NEC/Motorola (ie UMA/GAN) all had in common was a RAN gateway.

From a UMA perspective, one important outcome of the HNB specification is a validation of the RAN gateway architecture operators and vendors alike.

UMA/GAN is, of course, the original RAN gateway architecture.

I harken back to the early days of dual-mode handsets (DMH). The debate raged: IMS/SIP/VCC vs UMA/GAN. It’s taken 3 years for this to sort itself out in the market. Today, it’s clear that UMA/GAN and the RAN gateway architecture is the (only?) choice.

In the last 12 months, the debate started again, this time for femtocells. But just as quickly, a winner was declared: RAN gateway for HNB (a la GAN).

3GPP Selects Femtocell Standard, and it’s not (exactly) UMA/GAN

It was announced by multiple sources this week that 3GPP selected a reference architecture for femtocell-to-core network connectivity. This isn’t the completed standard, but sets the framework for the 3GPP RAN 3 working group to develop a full specification. The plan of record, per the 3GPP, is for the full stage 2 and stage 3 documents to be completed by Dec 2008 (in Release 8).

As has also been point out recently, the existing 3GPP UMA/GAN specification was one of a few architectures being considered as the basis for a formal 3GPP femtocell standard. The net result is that the UMA/GAN standard, as is, was not chosen. Sad but true. In order to reach an industry consensus on a HNB architecture there was a measure of compromise from all the participating companies. That’s the nature of standards: compromise.

However, while the HNB reference architecture is not full UMA/GAN, it adopts a number of key principles first introduced in the 3GPP GAN standard. For example, at the highest level, it follows an access-based (rather than core-based) approach, leveraging the existing Iu-cs/Iu-ps interfaces into the core service network. It also identifies the use of a specific protocol for solving a number of challenges associated with the ad-hoc deployment of devices over the internet.

Frankly, while the UMA community would have been delighted if 3GPP had adopted GAN lock-stock-and barrel (as they say in the US) as the formal femtocell standard, the likelihood has always been low. Fortunately, the agreed femtocell architecture builds on several core principles of GAN, which will help accelerate the market.

ThinkFemtocell has a good breakdown of different elements of the work.

For at least a little while longer, UMA/GAN remains the only 3GPP standard which supports femtocells today...