LTE: losing its voice?

One of the latest signs that LTE is moving into the "reality" stage of the hype curve is that equipment contracts are starting to pile up. As we went to press, LTE contracts had been awarded in Japan, Norway, Singapore, Spain, Sweden and the US, with Alcatel-Lucent and Ericsson the early winners, while in recent weeks, Chinese vendors Huawei Technologies and ZTE have been picking up momentum with LTE trial wins with Netcom of Norway and Telefonica in Spain, respectively.

Another sign that the hype bubble may be cresting is that some mobile players are already worried about how well LTE will play with legacy services - namely, mobile's two most reliably consistent cash cows: voice and SMS.

While most of the presentations, white papers and seminars on LTE in the past year have talked up the technology's multi-megabit data capabilities and the possible business cases that could be built on that, far less attention has been paid to LTE's ability to handle voice calls, apart from the fact that - unlike its 2G and 3G predecessors - it's not an old-school circuit-switched network retrofitted for packet traffic, but the other way around.

The assumption has always been that LTE would be rolled out as an overlay to existing 2G/3G systems, which meant that LTE could get in with the business of moving packets and leave the circuit-switched voice traffic to the old networks that are not only optimized for it anyway, but also aren't going anywhere. The other assumption was that the 3GPP's IP Multimedia Subsystem (IMS) platform would enable cellcos that wanted voice and SMS services on their LTE networks to duplicate them on the new platform.

However, T-Mobile International is not convinced that LTE will ship 100% voice-ready when it becomes commercially available later this year. It's banking on LTE supporting voice and SMS partially because it wants to launch with a full-service LTE portfolio and also because of the efficiency gains of running voice and data on the same network, which reduces the cost-per-bit (a key LTE selling point).

As such, T-Mobile has been rallying the LTE vendor space to revisit the voice issue to make sure it is ready. At February's Mobile World Congress, Emin GŸrdenli, T-Mobile International's SVP of radio networks (and technology director for T-Mobile UK), said his company wants voice capability in LTE "ideally from dayone".

A month later, T-Mobile spearheaded the formation of the VoLGA (Voice over LTE via Generic Access) Forum, with Alcatel-Lucent, Ericsson, Huawei, Kineto Wireless, LG, Motorola, Nortel Networks, Samsung, Starent and ZTE onboard to define specs for circuit switched voice and SMS over LTE using the 3GPP's Generic Access Network (GAN) standard. The forum already has published its stage 1 specs and stage 2 draft specs, which it has also submitted to the 3GPP.

However, as is always the case with fledgling technology standards, not everyone is convinced that LTE's voice-SMS capability is an urgent issue - and most of the skeptics are cellcos that are more sold on LTE's data-only proposition.

Fallbacks and stopgaps

One of the problems with running voice over LTE is that there are literally at least a dozen ways to do it, according to Ericsson CTO Michael Lee, which is why the 3GPP is focusing on two technologies: IMS and Circuit-Switched (CS) Fallback.

"You have to keep it narrowed down because it's asking a lot for device vendors to support a dozen different voice clients, and it also makes roaming a real problem," Lee says.

CS Fallback - which as the name implies involves pushing the handset from the LTE network to the 2G/3G network when the user makes or receives a call - is not only supported by handset vendors and roams easily since 2G/3G systems have agreements in place, but is also the most efficient use of network resources, Lee says.

"Both 2G and 3G are already there and are optimized for voice, so why design that capability in a whole new network?" Lee says. "Operators will maintain those networks, and at first you'll have only have islands of LTE coverage, so you need something to fall back on anyway."

The VoLGA group, however, maintains that CS Fallback is "stopgap" solution that is flawed in a number of ways - first and foremost being that it introduces latency issues for call setup that will hurt service quality.

"Switching networks takes time, and that's added time the user is waiting to make or receive a phone call," Steve Shaw, VP of corporate marketing for Kineto Wireless, writes on his Voice and SMS Over LTE blog.

What's worse, says Franz Seiser, head of core network architecture for T-Mobile International, is that neither CS Fallback nor IMS currently support SMS.

"We don't have one single mobile broadband proposition that is not using SMS," Seiser said during a presentation at the LTE World Summit in Berlin in May. "I don't mean user messaging, because not too many people will send SMSs from PCs, but service messaging that we use to serve our customers."

He cited examples such as "bill-shock" messages sent to users approaching a designated usage cap and notification messages sent to roaming users directing them to preferred roaming partners or letting them know how to make cheaper IDD calls.

"We did our own internal investigation and I was really shocked by how many systems will be affected if we don't support SMS anymore," Seiser said, saying "much more than ten systems" would have to be changed if SMS is not available in LTE.

VoLGA display

Meanwhile, the VoLGA Forum is also critical of LTE's reliance on IMS to support voice services in the longer term. Kineto argues in a VoLGA white paper that interest in IMS has waned as cellcos rethink their data strategies around "real" mobile internet access to popular sites like Facebook and YouTube, and those still keen on IMS want to use it for developing new services such as those specified in the GSM Association's Rich Communication Suite (RCS) rather than spend time duplicating existing services.

The VoLGA solution is to leverage the 3GPP-vetted GAN standard (which, it should be noted, is based on Kineto's UMA technology for home-zone services that blend 3G and Wi-Fi). GAN was initially designed to link mobile services to fixed-line broadband environments, but under last year's Release 8 for LTE, GAN can also support 2G/3G core network interfaces.

Put simply, says Kineto's Shaw, VoLGA uses a simplified version of the GAN spec to link the 2G/3G core - and their respective BSS/OSS platforms - to the LTE evolved packet core without resorting to IPsec tunneling normally used in UMA/GAN deployments and without requiring an MSC upgrade.

It's worth adding that while the VoLGA Forum is critical of IMS as a mechanism for supporting voice and SMS - not least because not many cellcos have adopted IMS in the first place - it's not positioning VoLGA as a replacement but rather a supplement that will work alongside IMS, handling the voice and text messaging aspects and freeing up IMS to handle RCS and whatever other new services the mobile sector wants to build.

Voice not a priority

VoLGA's claims haven't been publicly put to the test yet, so it's too soon to tell how effective a solution it is compared to CS Fallback. The only chief criticism from the supply side has been from Nokia Siemens Networks, which is not part of the VoLGA Forum and sees the technology as a solution in search of a problem.

"LTE is for high-speed data. If you want to support voice and you have a legacy network, you can use CS Fallback, or if you don't have that but you have an LTE license, you can support voice through IMS," says Mike Wang, NSN's GM for Taiwan, Hong Kong and Macao. "It is not necessary to implement VoLGA to support it - the ability is already there."

(Note: Since that statement NSN has struck a deal to take over the LTE business of VoLGA supporter Nortel, but hasn't announced any new policy or plan regarding VoLGA.)

But even if VoLGA works as advertised, the tough part may be convincing LTE licensees they need it at all.

Indeed, T-Mobile is the only operator involved in VoLGA thus far, and the only major cellco to publicly emphasize the importance of legacy voice/SMS support in LTE.

The division was clearly spelled out at LTE World Summit. T-Mobile's Seiser made the case for VoLGA and the urgency for LTE to fully support voice and SMS - and got some support from 3 Group technology director Ed Candy, who said he wouldn't support any technology with "fundamental holes" that limit service options, according to Unstrung. But Marc Fossier, executive VP of corporate and social responsibility at Orange (and former CTO of France Telecom), said that LTE will serve their needs just fine as a data-only platform for the time being, and that CS Fallback, while "not nice", could still support legacy voice/SMS.

Even Lee from Ericsson - a member of the VoLGA Forum - says that some of the operators he's talked to see LTE as data-only. "Some don't even bother with voice - mobile broadband is the only app," he says. "The main devices initially will be notebooks and netbooks and dongles, which don't need voice support."

Marc Einstein, industry manager for the Asia-Pacific ICT practice at Frost & Sullivan, agrees. "When you talk to the major operators about LTE voice strategies, almost all of them say there's no need for backward compatibility from a device perspective," he says. "Devices for LTE at DoCoMo will be LTE-only. Verizon is not planning to have a dual CDMA/LTE device. What I get from that is that LTE is going to be a data-only play initially."

One operator that sees LTE as a data-only technology at first is Hong Kong CSL, which acquired its LTE spectrum earlier this year. The reason, says CSL CTO Christian Daigneault, is that mobile data is where the growth is.

"In Hong Kong, mobile data is increasing, but voice isn't - it's relatively steady," he says. "So we are focused on using our LTE spectrum only for data because we need the capacity."

To be sure, Daigneault adds, he does see value in running voice on LTE in terms of the added spectral efficiency gains.

"Running voice over LTE will reduce the cost per bit, so when we start reaching full capacity, we can look at doing voice over LTE, as well as increasing the efficiency of our 3G spectrum, doing things like voice over HSPA," he says. "But that is maybe three or four years down the road for us."

Skype-proofing LTE

That said, it's still worth hammering out the details on voice/SMS support now rather than wait for cellcos to need it, says Lee - and not just because of the spectral efficiency gains and cost-per-bit savings.

"Once voice comes onto LTE, that will be an opportunity for operators to differentiate themselves from their 3G offerings," he says. "There's no point in offering the same voice apps that customers get with 3G. They should focus on IMS and the capabilities it can bring end-users - like tying text chat, videoconferencing, file transfer and photo sharing into calls."

That will be particularly crucial in competing against over-the-top services such as Skype and Google Voice, he adds.

"When fixed-line operators launched broadband, they opened the market for over-the-top apps like Google, Skype. Mobile operators don't want to repeat the same mistake," he says.

EV-DO also weighs LTE voice options

At first glance, the debate over LTE's ability to support voice services may seem primarily aimed at GSM/UMTS operators, but CDMA operators like Verizon Wireless and KDDI which are planning to launch LTE late next year face a similar dilemma.

At the CDMA World Forum in June, Richard Brown, VP of Alcatel-Lucent's CDMA product unit, had plenty to say about voice strategies for CDMA operators planning to adopt LTE, and CS Fallback will definitely be an option.

"If you have a device that can support both 1x as well as EVDO and LTE, users can make voice calls by falling back to the 1x network," Brown said in a presentation.

Furthermore, he added, the LTE/1x EV-DO device would be able to switch to the 1x network even in the middle of an LTE data session, and then resume that session once the call is finished.

"With that capability in place, we can continue to support 1x voice for a very long time," Brown said.

He also took the opportunity to talk up 1x Advanced, a Qualcomm-led program to boost the voice capacity of existing 1x network by four times the current level - which in turn would free up spectrum to add more capacity to EV-DO and LTE services.

That said, Brown added that the business case for running voice on or off the LTE network will vary from one operator to the next.

"Are you thinking of using LTE just for data and keeping voice under various networks? Or you are thinking about using VoIP on EV-DO and LTE? Either way, there are a number of factors to consider," he said.

Such factors include the importance of voice and/or all-IP to your business model, capex priorities, available capacity, roaming arrangements and handset complexity, he said.

- John Tanner and Fiona Chau