Gordon Mansfield is responsible for Small Cells at AT&T and has also been Chairman of the Small Cell Forum since last September. We asked him if there really is a mobile data traffic problem or not, and how urgently small cells are needed to address the problem. He also shared his views on LTE adoption, new features such as VoLTE, and some of the LTE-Advanced features which can't easily be supported by small cells.
These insights are very timely in the run-up to next week's LTE World Summit.
Is there really a mobile data tsunami or not? Is the future growth of mobile data traffic overestimated?
I'd point out that a lot of European countries haven't fully launched LTE yet. AT&T has publicly stated traffic growth figures which clearly show that when LTE was launched, bandwidth demand takes another step up. The extra spectrum deployed for LTE helps, but as users move over we've seen that LTE capacity is exhausted at a much faster clip. Even with the improved spectral efficiency, LTE alone doesn't offset demand for extra capacity. Some will say that in these cases, they start to see a lot of capacity driven use cases for small cells in this segment.
Worldwide, I think the cellular industry will move to LTE a lot faster than it did with other technologies. In some countries, the transition to UMTS is still going on today, Europe is not alone. In LTE we can see North America is moving very fast and Asia is doing the same. We think Europe will go quickly once LTE has started.
Won't the large installed base of 3G smartphones restrict take-up of LTE?
When 3G was launched around 2002, it took time for 3G smartphones to make full use of the capabilities. The device mix when launched wasn't that different from GSM. It was perhaps 2005 before we saw the first truly useful devices and not until 2008 that a full ecosystem had developed.
For LTE, the end-user devices are already very good and make full use of the higher data speeds. These serve not just high-end categories but we can see the range of mid-tier devices is filling up too. The only thing left that hasn't shifted rapidly to LTE is the lowest tier of devices. When visiting Seoul, Korea, I saw the same situation and this is also the case in Japan. A lot of the latest phones in Europe being sold today are already LTE capable too.
The wide variety of frequency band plans in use for LTE affects the true international reach of a single device. I can assure you that there are a lot of conversations happening between operators internationally to rationalize which bands are relatively ubiquitous worldwide. I'd expect each continent to adopt a couple of frequencies that become international roaming bands and facilitate worldwide LTE roaming.
Is VoLTE (Voice over LTE) ready for widespread commercial deployment, and how will it affect takeup of LTE small cells?
VoLTE (Voice over LTE) standards are now getting to the point where they are relatively complete. There were some features/functionality to be put in place to make it compelling, and this is now done. All three Korean operators have launched it on a wide scale. I'd say that 2013 will be all about testing and getting things in place. 2014 will see a wide ecosystem supporting VoLTE. Once that happens, then all LTE networks will be VoLTE enabled from the start.
This will start to drive small cell capacity and link budget, especially uplink, which has to be paid particular attention to. It will require a fairly dense layer of network in order to support the good quality voice service that people have grown accustomed to. In turn, this will drive further densification and greater adoption of small cells.
Korea, which initially launched LTE as a data only service has offered VoLTE since the end of 2012 across all operators. During the Forum Plenary in March, we visited a couple of operators both from a Small Cells and a VoLTE perspective. They were in the early stages of VoLTE capabilities with HD Voice, and there are more necessary enhancements still to come, but early learnings identify where and how to make adjustments.
As VoLTE becomes more mainstream across the device portfolio, the cost of individual devices supporting an ever wider range of different frequencies and modes becomes more expensive. As the world comes together in one technology – and we can start to see the industry on a global basis moving towards LTE – at some point it makes sense to reduce cost point down by removing technologies. We can't do that until the LTE network has full capabilities.
Voice is table stakes. Moving a significant proportion of voice capacity across to LTE is going to need some level of densification to maintain performance. Not every user will be on a small cell all the time. We would certainly need to have traffic steering based on multiple conditions, e.g. a 50 mph mobile user won't be moving quickly between small cells, but pedestrian traffic walking slowly would be better served by small cells.
What's VoLTE like to use in practice? Is it any different from 2G or 3G voice?
There is a very fast call setup time - blazingly fast compared to GSM or UMTS, but what really impressed me was HiDef (High Definition) voice. I've spoken with people who've used it for a while. It seems that having HiDef voice initially seems to make little difference – you could take it or leave it. What's interesting is that after a user had it for some time, then they really noticed when not using it anymore.
I believe that making investments to introduce HiDef voice over VoLTE is a better investment than legacy upgrades, but this choice will vary from region to region depending on deployments and plans. I think it will be a key driver for VoLTE adoption.
Are some of the most demanding LTE-Advanced HetNet features included in small cells?
Today I would not include the most aggressive CoMP (Co-ordinated Multi-Path) features within Small Cell Forum work, although I wouldn't rule it out in future. Some features like CoMP would be very difficult to deploy using integrated small cells because of backhaul transport limitations. IP over Ethernet doesn't meet the low latency requirement – dark fiber or similar connections are needed.
I think some parts of the aggressive (downlink) features of CoMP will be deployed in isolated use cases, not on a large scale, until the industry comes up with solutions to support it. Where alternatives are difficult and costly to implement such as football/soccer stadiums, CoMP makes a lot of sense there, but it could be very costly if broadly deployed. There should be a better return on investment from mainstream capabilities existing today.
There should be a lot of benefit in aspects of uplink CoMP and eICIC (enhanced Inter-Cell Interference Co-ordination) that don't need such high latency demands. This gets you significant enhancements and helps justify the business case to deploy on a wide scale.