GAN for LTE: A shot accross the bow for 3GPP?

LTE is a great technology but it has one Achilles heel: No built in telephony solution that will work when network are first rolled out. I've blogged before about the different alternatives for voice over LTE before, such as IMS, Voice Call Continuity, CS Fallback and a Study Item how to connect a circuit switched MSC to the LTE radio access network. While the last one is only a study item and not standardized yet, the first three are standardized but either complicated or, in the case of CS fallback, not a long term solution.

Looks like some operators are not very happy with the state of affairs and have looked elsewhere for help. At the recent LTE World Summit in London, Motorola and China Mobile have both said that they are considering GAN over LTE as a further alternative. GAN is used today in some 2G/Wi-Fi phones for Voice over Wifi at home. What it basically does is it establishes an encrypted IP tunnel between the mobile and the GAN controller at the border of the operators core network, and then tunnels both signaling and speech data to the MSC. Furthermore, the mobile and GANC are able to perform handovers between the Wi-Fi connection and GSM for ongoing calls. In effect, GAN hides the access network (Wi-Fi in this case) from the operators circuit switched core network.

The solution proposed by Motorola at the conference, GAN would do the same for LTE. After all, an IP connection is an IP connection, the core functionality of the GANC does not care if the interface to the user is Wi-Fi or LTE.

On his blog, Zahid has a slide of the Motorola presentation that shows how the network setup could look like. The figure shows an interface between the Mobility Management Entitiy (MME) and the GANC which I think is probably for handover co-ordination. When the MME detects that a handover is necessary to UMTS or GSM, it informs the GANC which can then ensure that an ongoing voice call survives the handover.

The advantage of this approach over the evolved MSC approach, which only exists as a 3GPP study item so far, is that the current circuit switched network infrastructure does not have to be changed. Even though the changes required for the eMSC are minor in comparison to IMS and VCC, they still need to be standardized and implemented by the major infrastructure manufacturers. The GAN approach on the other hand requires some modifications in the mobile and the GANC only, which could well be made without waiting for 3GPP.

>From a design point of view I think the eMSC would be a much more elegant and native approach to the issue. But in practice, elegance does not always win if it is more complex, or as in this case, meets reluctance in the standards bodies.

So, let's see if this 'dear 3GPP, we either do it with you or without you' announcement will change the attitude of some players.

Thanks to Zahid Ghadialy for his coverage of the LTE World Summit