SGSN and LTE

With LTE gaining more light I guess most of the network manufacturers are ready with basic attach, default and dedicated bearers, X2 and S1 handovers. But LTE as standalone will not make sense as it will deployed in just few places. Rest of the world will be still running on 2G/3G/CDMA networks. So there is a strong need for 3G/CDMA convergence with LTE which means inter RAT handovers. I think Europe will see 3G convergence with LTE while Verizon is working hard to converge their CDMA network into LTE in US. So, may be, this year will be about convergence and figuring out a proper solution for voice over LTE.

My interest here is SGSN. SGSN is a major signaling element in 3G packet core. It runs RANAP/NAS protocols towards RNC and GTPv1 towards GGSN. If 3G is to be converged into LTE then there has to be a software upgrade on SGSN as it needs to speak GTPv2 over S4 interface with LTE MME. This makes SGSN much more important and signaling intensive device. I am sure that both operators and equipment manufacturers are looking closely at the SGSN's now. The question here is how will the operators choose which 3G networks should support LTE and which shouldnt. Imagine this. I have a UE which has APN configured with a LTE PGW. While in roaming I may not find a LTE network so I will have to switch back to available 3G network. Now, that 3G network should be capable enough to support my UE's existence in the network. This means SGSN should go all the way to talk to LTE MME, pull the bearer contexts, map the QoS and serve the UE. This needs quite a lot of processing. Will existing SGSN be capable to do all this? This will probably need much powerful hardware on SGSN. I am very keen to see how things will go in this direction. I am hoping that Mobile World Congress this year will have some answers. Any thoughts?