There's a lot of speculation and misinformation regarding how voice service will be delivered once LTE is deployed on a large scale. Voice service may seem trivial but when some 3G technologies first rolled out voice was problematic. Subscribers weren’t happy when fancy new phones with upgraded service plans couldn’t make a phone call. In reality, the way North American operators are deploying voice service is a little surprising.
For a long time, any discussion of LTE was so focused on data that we hardly ever mentioned voice. In fact, we as an industry were very much unsure how voice would eventually be delivered. Proposals and ideas came from different camps, all of them in seeming opposition to the others.
The most intuitive method, described in the 3GPP specification 23.272, is circuit-switched fallback (CSFB), which basically hands voice calls over to legacy networks. While it makes sense to use networks that already exist, testing shows that the call setup can take up to seven seconds in some scenarios. Most subscribers would have an issue with setup times like that.
The next big idea came as an industry specification released in September 2009. Voice over LTE using Generic Access (VoLGA) grew out of an existing technology called the Generic Access Network (GAN). Where GAN defines seamless handovers between cellular and 802.11 (WiFi) networks, VoLGA replaces WiFi with a secure tunnel on an LTE bearer. VoLGA had a lot of support, having been endorsed by such heavyweights as Alcatel-Lucent, Huawei, Deutsche Telekom and Ericsson.
VoLGA seemed to be a cost-efficient solution since it did not require an IMS network. However, many operators had other reasons to deploy IMS, so this was a sunk cost, and VoLGA was not as cost-efficient as some had claimed.
Still, VoLGA had the benefit of being the only clearly-defined direction supported by industry leaders. That changed drastically on November 5, 2009 when a consortium including Verizon Wireless, AT&T, Alcatel-Lucent, Ericsson, Vodafone and other key players announced the "One Voice" initiative.
One Voice is neither a standard nor a specification. It is an IMS-based profile built with existing 3GPP-defined components. To seemingly seal the deal, the GSMA announced its support of “One Voice” during this year’s Mobile World Congress by publicly announcing the voice over LTE (VoLTE) initiative. Note that VoLTE is not just an endorsement of One Voice. VoLTE has already added some clarity in the area of roaming. More importantly, VoLTE is the GSMA’s commitment to further defining how voice and SMS will work in the future.
So what are the North American operators doing? Here’s a hint: What does it cost AT&T or Verizon Wireless to deploy 2G/3G service? Answer: Nothing… it’s already there. So they’re using CSFB? Not quite as the latency was not acceptable. The solution is to use devices with two radios- essentially a voice phone and data terminal within a single enclosure. Of course, this is nowhere near as simple as it sounds. It requires a massive amount of testing to ensure that neither technology interferes with the other at any layer of the protocol stack. But it is what we'll see in a few months and so far it seems to be an effective and efficient solution.
Michael Keeley is a director of product management within Spirent Communications’ wireless test equipment division. He has led various teams involved in wireless network emulation and automated systems used for testing mobile devices. Prior to joining Spirent in 2000, Keeley worked for Lucent Technologies. He received his BSEE and MEng from Cornell University.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.