A useful and worthwhile event, sharing progress, complexity and the underlying business value of voice and video services on LTE – both VoLTE and so-called over-the-top services. My perspective was investigating how this relates to adoption of LTE only networks and small cells including neutral host. My findings may surprise you.
Current status of VoLTE
While LTE has been a great success since launch in 2011 and with over 700 live networks today, VoLTE has been slow to catch up. Diane Myers, IHS, noted 41commercially live VoLTE networks and 19 Voice over Wi-Fi (there is some overlap between both). It’s principally a voice service, with Video over LTE (ViLTE) and Rich Communication Services (RCS) still relatively rare.
Operator Case Study of VoLTE deployment
Phil Sheppard, Director of Network Strategy at Three UK, gave an excellent keynote case study of their VoLTE deployment. He was very clear that the business driver was entirely around improving in-building coverage. They had acquired 10MHz of 800MHz spectrum in the LTE auction and use this exclusively to extend coverage from their existing macrocells into areas and inside buildings which 3G at 2100MHz couldn’t reach.
The three challenges were
- Compatible devices (probably the biggest difficulty)
- Implementing LTE radio coverage at their cellsites
- Installing and commissioning an EPC/IMS Core network
The timescale was:
- Feb 2013: Spectrum auction concluded
- May 2014: Voice over Wi-Fi was introduced using a standalone App.
- Sep 2015: VoLTE added
- Sep 2016: Embedded Wi-Fi Calling
It was a huge project, the scope far exceeding original expectations and affecting almost every part of their operations. They spent 22,245 man days (excluding vendor staff time) – more than 100 man years of effort – mostly involved with running huge numbers of test cases on the network and devices.
Focussing on the primary business driver led to some surprising decisions. Only VoLTE capable phones were allowed to access the 800MHz band – this avoided the scenario I’ve heard of elsewhere, where non-VoLTE devices fallback to 3G only to find they can’t receive voice calls in areas with poorer 3G coverage.
VoLTE has only been deployed in the 800MHz band and not yet in their 1800MHz LTE spectrum.
There had been very careful constraints on which devices were supported, with Android and certain iPhone models carefully introduced. In response to my question, Phil confirmed that inbound roamers with compatible handsets would benefit from VoLTE (although the roaming interconnection would use traditional 3G voice signalling/billing).
The outcome has been very positive – an increase from 18 to 23% in Net Promoter Score, and 34% reduction in churn. Call success rates are very high. Few customers may appreciate or realise their calls are made via VoLTE – the service operates exactly as for 2G or 3G. HD voice quality for intra-network VoLTE calls would be better, but that only occurs when both parties are using the 800MHz band.
Bryn Jones, Three’s CTO, explained their longer term plan includes deploying VoLTE in other frequency bands and refarming 3G spectrum to LTE. He thought that operators were generally behind the curve on advanced voice services compared with Facetime, Facebook etc. but isn’t protectionist about their customers using OTT services. He thought that as more operators deploy VoLTE and video calling, so that it becomes just another seamless service, it may win back use from Apple, Facebook which are limited to their own closed user groups. It’s a potential bonus that doesn’t factor into their own business case for VoLTE deployment.
So will VoLTE be easier to deploy in other networks?
Have the “pioneering” VoLTE first adopters overcome most of the issues for VoLTE rollout? Clearly there have been numerous issues with the first round of compatible smartphones/devices, and I’d expect most of those to have been resolved. But several test vendors at the event seemed to be rubbing their hands with glee at the thought of the extensive testing process which many operators will need to go through in the coming months ahead.
Rhode and Schwarz, perhaps best known for their low level (i.e. physical layer) test gear, assured me that they saw plenty of issues to keep their customers busy. EXFO, Emprix, TEMS/Infovista and others explain how test and performance monitoring needs to be concurrently tracked at several protocol levels, to identify the underlying causes. Others mentioned to me that device compatibility remains an issue.
However, I think it would be reasonable to assume that the bulk of the really difficult technical issues have been overcome by now. What will remain for each new VoLTE rollout will be replicating and ensuring services operate exactly as for 2G/3G, and this will vary widely depending on legacy network service implementation.
I also think is quite likely that we’ll never reach 100% VoLTE rollout across all LTE networks. I’ve seen earlier analyst estimates that perhaps only 65% of all networks would eventually adopt it. For smaller operators with adequate 2G/3G coverage today, the business driver may be far less urgent. On the other hand, for greenfield networks such as India’s Reliance Jio, it’s a no-brainer.
Handset costs are a factor but I was assured that prices for low end 3G smartphones are down to $50 and LTE-only $100.
Proponents of Wi-Fi Calling highlight that this remains an alternative cost effective method to reach poorly served homes, especially in lower cost countries. Frederick Reynolds, Taqua’s VP Marketing, points out there are plenty of low cost Android devices which could easily be served by Wi-Fi Calling. They have implemented several smart ways to provide seamless Wi-Fi to 3G handoff and still see it as a viable alternative. (Ed Note: Taqua was recently acquired by Sonus last month).
I’d argue that Wi-Fi Calling can cause more problems than it solves in public venues or larger buildings, but may have a market in homes with adequate wireline broadband or where low cost is critical. (See our recent write-up of a case study where Enterprise Wi-Fi Calling caused more issues than it solved).
The value of HD Voice
Despite their best attempts, I found little commercial enthusiasm for the higher voice quality that VoLTE provides – sub-second call setup being the other. The codec that VoLTE uses for this higher voice quality can also be used on 3G networks (e.g. as Sprint does). It’s generally only relevant on intranet (Mobile to Mobile) calls today.
I don’t know of any network that charges a higher rate for the premium quality.
A strong champion of HD voice was Rod Randall, Sirus Capital Group. He notes that PGi is the third largest audio conference network worldwide today, and sees huge value if each mobile network would interconnect to their local equipment via direct HD. The quality of the conference audio is so much improved that he expects users would actively choose to “dial in” using their mobile. You wouldn’t need to use a specific App, just the embedded voice function in your smartphone.
HD Interconnect may be some way off
Sidebar discussions with senior architects in several operators reveals that implementing end-to-end VoLTE calls between networks may not be nearly as easy as you might think. It seems that many voice calls are quickly converted to the common landline G.711 codec format, where standard (i.e. non codec specific) services can be applied. This can be anything from audio announcements, ring tones, voicemail to (my speculation) lawful intercept. Handling multiple codecs for each of these services would add cost and complexity.
In any case, the vast majority of calls are intra network (typically over 70%) and those to landlines wouldn’t benefit. What we might expect to see is each multinational mobile network group implement end-to-end VoLTE between their national subsidiaries first. Networks with both fixed and mobile arms, such as BT UK, KPN Netherlands, could potentially implement HD voice between fixed/mobile calls but I don’t see a strong commercial demand as yet.
I wasn’t aware there are at least three different ways of doing this:
- Use existing 3G roaming methods. As explained above, inbound roamers may be served using VoLTE by the visited network but this would be presented to the home network as though it were just a regular 3G roaming call. It’s the easiest/simplest/quickest solution today.
- LBO (Local Breakout). The home network IMS delegates call control to the visited network, which actively manages the call. Very similar to 3G, except the call processing is managed by IMS rather than a 3G switch.
- S8HR (Home Routing). The visited network acts like a dumb data pipe, with the VoLTE call being actively managed and controlled as if at home. Priority is given to ensure adequate voice quality.
It seemed to me that this remains an area of high complexity and confusion. The speaker from NTT DoCoMo explained why they had chosen S8HR, and the resulting simplicity. Questions from the audience noted that the visiting network would miss out on call charges, and queried how unusual services such as emergency calling, lawful intercept etc. would be handled.
iBasis, which handles roaming interconnect, prefers the S8HR option. They’ve seen over 82 VoLTE domestic launches but limited VoLTE roaming, which has mainly been S8HR to date. Around 80% of roaming minutes are for calls back to home. Security breaches are a risk, and one example given was where the proxy CSCF ip address was manipulated to bypass the PSCF on an Android phone.
OTT Voice and Video services
Competition for higher quality and wider reaching services comes from OTT applications such as Facetime, Facebook Messenger, Microsoft Skype and Google Duo.
One architect suggested that they could choose to configure their network to support these services with the same quality as VoLTE – it just needs the quality/priority to be set appropriately.
Summary and Implications
Clearly, VoLTE adoption today is being driven for commercial reasons other than improving voice quality or adding new services. Verizon Wireless, China Mobile, EE UK and others are implementing it as a “necessary evil” to access new LTE spectrum and greater LTE data efficiency and speed. Other networks may not have such pressing need, and so the pace of VoLTE adoption may remain fairly slow.
There is substantial complexity of implementing VoLTE to a high quality and some lower cost/smaller networks may choose to remain with 3G voice (CSFB) for some time. For those that do move forward, test and performance products and services will be essential. Expertise learnt from pioneering networks will be hugely valuable.
VoLTE interconnect and roaming isn’t a strong business priority, however it’s a welcome surprise that visitors with VoLTE capable smartphones appear to be able use that feature even where their home network doesn’t support it.
The conference provided a very helpful checkpoint for VoLTE status and progress in the real world, enabled some good networking with senior architects involved in today’s implementations, and highlighted issues still to be overcome.
For me, this makes 3rd party LTE only small cell networks more viable more quickly but I’d say that today’s requirement remains both 3G/LTE (for device compatibility). In the medium to longer term, VoLTE will dominate and implicitly allow a simplification to LTE only.