Moving to an all-IP network is a necessity

There's little argument that mobile carriers need to move to all-IP networks if for no reason than to offload the abundance of data traffic that's now almost matching voice on their 3G networks. On the other hand, carriers are reluctant to make the move to IP because, as stated, data is almost matching voice and voice-over-IP is problematic.

The best way to accomplish both--offload data and maintain a strong voice network-- is to compromise now with the introduction of IP as part of an LTE network and then add voice-over-IP later.

"There are plenty of reasons to start a full mobile IP transformation and move to all-IP sooner than later. There are already tons of IP-based applications moving across mobile networks and the unification of all mobile network protocols around IP (all-IP) actually enables the network operators to do something with it," said Aleksandar Pavlovic, senior manager of mobile marketing solutions for the IP Division of Alcatel-Lucent.

The move will be made in steps. The network will be IP because that's the most efficient way to offload the increased weight of data traffic. The applications will be a mix.

"You're going to have multiple technologies coexisting with each other and then targeted for various specific applications," said Godfrey Chua, research manager-wireless and mobile infrastructure at IDC. "It's actually going to be a long process before everything is running off an LTE network."

That doesn't mean operators aren't moving to IP now. They are, but "the minute that Verizon launches LTE doesn't mean that they're going to migrate their voice calls to LTE right away," Chua said. "It's actually going to be a long process before everything is running off an LTE network. Verizon, in our conversations with them, fully expect their 2G CDMA network to support voice until 2020."

The Evolved Packet Core makes the move more palatable by providing a migratory path to an all-IP future, Pavlovic said.

"You really need to have a very high performance purpose-built dedicated IP element to be able to cope with the increased processing and increased bandwidth at the data plane," he said. "You have to have a very highly capable product, highly reliable and have it conditioned for performance when all this additional processing comes into play."

Pavlovic said it would be a misstep to follow a path of revamping equipment built for 3G networks for the IP-centric 4G space. Instead, carriers should, and will, move to all-IP in steps that start with building IP network infrastructure.

"You need to have a new type of architecture to address these types of environments," he said.

Peter Jarich, principal analyst for wireless infrastructure at Current Analysis, believes that carriers are moving to all-IP and, as Pavlovic suggested, are doing it in steps.

"No one's saying when you move to LTE you have to keep your voice on that EPC from day one. There are solutions to leave the data on LTE but throw voice onto the 2G or 3G networks," Jarich said. "You can put your voice on your LTE network and do it as voice-over-IP and that's where everyone wants to go in the long term. In the near term, the question is how to get there."

Circuit Switched Fall Back (CSFB) can be used to continue reliable voice service while delivering data to devices such as dongles and netbooks that are gobbling up bandwidth.

"The first phase before you get to voice-over-IP is going to be CSFB so you'll just roll back to your 3G network in cases where a voice connection is required on an LTE device," said Phil Marshall, senior research fellow, technology strategies at the Yankee Group. "You're not building your LTE network as a voice network to begin with and the longer term strategy is to drive the voice-over-IP architecture."

T-Mobile, among all wireless carriers, is "pushing really hard" for a compromise called VoLGA (Voice over LTE via Generic Access) that breaks mobile voice and messaging apart from the IP-based LTE network and puts it on the existing 3GPP Generic Access Network (GAN).

That's something of a quick-hit solution that might be more troublesome than helpful, Chua said. "The process is always more complex than it seems [although conceptually it's a very straightforward concept," he said.

Marshall agreed. "The longer term strategy is to drive the voice-over-IP architecture as opposed to having an overlay UMA or VoLGA infrastructure," he said. "Many of the players are thinking of LTE initially as a data-only network."

To get to all-IP will require patience. "The reality is it's going to be different. It's not going to be Verizon saying it's launched an LTE network and now it's going to run all its voice over LTE," said Jarich. "It's a work in progress."