By Paul S. Cawte
Despite the introduction of LTE with its heavy focus on improved mobile broadband speeds, mobile operators still rely on voice and SMS services for a large part of their revenues (roughly 70% globally). In previous generations of mobile technology these services were explicitly supported as part of the mobile network stack, with voice bearers supported in the radio access network and SMS making use of the voice signalling mechanisms. Indeed in 2G mobile standards data was originally only supported by sending data over a nailed up voice channel (high-speed circuit switched data or HSCSD) with the native data transport mechanisms of the general packet radio service (GPRS) introduced later as part of the so called 2.5G standards and in enhanced form as EDGE with 2.75G.
With the advent of 3G mobile standards (such as UMTS) voice and data were catered for on an equal footing, with distinct voice and data air interfaces being defined, each optimised for their respective payload. The lu-CS interface, used for voice, uses circuit switching to provide deterministic, guaranteed capacity for in-progress voice calls. In contrast, the lu-PS interface, used for data, uses packet switching on a shared data carrier to maximise the efficiency of data transport.
Compared to 2G, the 4G LTE standard turns the situation on its head and is optimised for data services without any specific native support (i.e. circuit switched) for voice transport. The rational for this is that broadband traffic is now the predominant use of mobile bandwidth and it is better to optimise the network for data transport and carry voice as an application over data using voice over IP. This is possible in an LTE network due to the substantial enhancements in the data plane, which compared to 3G has significantly reduced latency and has the quality of service mechanisms required to support a good quality voice service.
Although LTE was, from the outset, designed to support voice service via voice over IP (VoIP), the call flows, their associated signalling and media encoding where not defined or standardised. So although all the “hooks” where in place to support voice their was no voice standard that an operator could deploy and indeed there were many options, a number of which were debated at length by equipment vendors and network operators as part of the standards process.
Given the importance of voice (and SMS) this lack of standardised support for voice could be regarded as a serious omission. To mitigate this a mechanism called circuit switched fall-back (CSFB) has been introduced which entails the phone switching to 3G operation when a voice call is to be made or received.
As indicated above, there has been much debate about the standard approach for supporting voice (and SMS) over LTE. One approach, called VoLGA, which was proposed was to utilise existing UMA mechanisms such as those supported on some mobile phones (notably Blackberries) to tunnel voice across WiFi. Whilst this would have leveraged existing voice switching networks it had the disadvantage of being quite backward looking and not providing a future path to full multi-media communications. Instead the approach using native VoIP with SIP signalling and an IMS (IP multi-media subsystem) core found favour as the approach for voice over LTE (VoLTE) with circuit switched fall-back being used as a transitionary step.
As of today (September 2013) VoLTE standardisation is more or less complete and both handsets and network equipment are available. Nevertheless, there are very few deployments of VoLTE – to this author’s understanding one in the US (MetroPCS) and three in South Korea (SK Telekom, KT and LG U+). Whilst it is clear that operators will eventually embrace VoLTE what is not clear is how rapidly operators will move to do so.
On the one hand aggressively introducing VoLTE could enable network operators to refresh their voice product set, so as not to be left behind by the over the top (OTT) voice and messaging service providers such as Skype and WhatsApp. On the other hand, existing voice and SMS revenues are not yet necessarily under threat and it may be better to take a slow and steady approach. This paper sets out the different approaches network operators could take and the rational for preferring one approach to another.