Avren's HetNet Europe conference, formerly known as the Bath Basestation Conference, attracts a fairly technical crowd who are dealing at the sharp end with the issues of mobile radio networks. A tremendous range of information is covered, both during the presentations and the offline networking, of which I've covered only a sample below. While few would dispute the growing trend of mobile data traffic (although some do and did), I sensed that most European operators seem set to deploy a combination of UMTS900, Wi-Fi offload and 3G small cells before embarking on heavy investment in LTE. From a technical viewpoint, today's biggest challenge appears to be figuring out cost-effective backhaul for public access outdoor small cells. Otherwise, we saw new business models and players entering the fray, offering to solve the logistical and practical problems of how to scale up and deploy large numbers of small cells when you don't have the CAPEX budget, staff or systems in place to do so.
So what is a HetNet?
The conference opened by asking a panel for a definition of a HetNet and why they are important.
The panel agreed that HetNets are a cellular radio network using a mixture of radio technologies (3G, LTE and Wi-Fi) all working together, typically in a localised area with tight integration providing mobile services with predetermined QoS policies. We can expect to see HetNets actively deployed within 2 years, said Remus Sepp of T-Mobile UK.
Erol Hepsaydir (3 UK) believes it's all about getting optimum performance out of the total system, using all options at an operator's disposal to satisfy customer demand.
With data constituting over 95% of all traffic on every (Western) European network today already, we need to optimise for data rather than voice. He believes traffic demand is only going to increase because a significant percentage of phones sold today are smartphones, increasing their percentage across the installed base. He's seen that it is smartphones, rather than data dongles, which dominates the data volume. This drives the need for HetNets to make best use from all available aspects of mobile networks.
One topic which wasn't clearly resolved relates to Wi-Fi. Some delegates treat Wi-Fi as a single option, while others differentiated between private use (e.g. on your home/office Wi-Fi) and public access (i.e. through a carrier/service provider). Few seemed to know or understand how much of their customer's traffic is offloaded to Wi-Fi. I felt there wasn't enough discussion about how (carrier) Wi-Fi would be tightly integrated and seamlessly interoperate with the traditional 3G/LTE cellular service.
What are operator's expectations from their HetNets?
Steve Brown from Telefonica O2 UK painted a picture of what their network might look like by 2015. He believes that traffic won't be equally distributed, so there will be very localised hotspot areas which will require as much as 100x traffic carrying capacity of today, while other areas might be adequately served from their existing macrocell sites. He predicts that high traffic areas will be using 4x 3G HSPA+ carriers (2 at 900MHz, and 2 at 2100MHz), with 7 macrocells per square kilometre complemented by several indoor and outdoor small cells.
During their initial trials with public access small cells, they found that co-locating them with 2G GSM microcells wasn't effective – people don't consume data in the same places as they do voice calls. They also found they needed higher power outdoor small cells (2 to 5W RF power) to overcome the dominant macrocell signal.
TIM Italy also expect to be using the full capabilities of available spectrum, with their longer term goal including extensive use of LTE @ 800MHz in most places, allowing them to reach more remote regions than today, complemented by LTE @ 1800 and 2600 for small cells in urban areas to provide capacity. Strict regulatory limits on RF power prevent them from expanding capacity on their macrocells as much as other operators, driving the need for small cells more urgently.
Andy Dunkin from Vodafone group described their OneSON vision, which aims to maximise the performance from combined 3G/LTE and Wi-Fi access. They will measure their success with measures of both customer experience (seamless user experience) and efficiency (maximising performance, QoS and operational effectiveness).
Achieving this will require substantial change in four core areas of their operations/engineering teams, affecting Planning, Deployment, Maintenance and Optimisation. Using the same number of staff to handle the greater complexity and scale, he believes considerable automation of many of today's manual tasks will be essential.
It's not just the products that are required to build out HetNets, processes need to change too
Mike Schabel of Alcatel-Lucent developed this theme. He believes the industry is really just embarking on a journey which involves rethinking and retooling how networks are built and deployed. Activities which work well today for macrocell deployment will need radical change to adapt to the needs of small cell and HetNets. Expounding on a wide range of factors from site acquisition, network optimisation to mass deployment with less skilled staff, he explained how these will need to change. These will differ by country and operator – there is no single blueprint - and will evolve as operators gain experience.
One example related to site selection. This is no longer just about optimal RF positioning. The lowest cost option may be the least intuitive answer. He's seen a range of 10x of cost between different options for the same site. There is a need to do full TCO analysis of all options, and keep an open mind as to which might be best.
More spectrum and increased spectral efficiency will not be enough to meet projected data demand, driving the requirement for small cells. He believes the scramble for (small cell) sites has already begun. Time is of the essence.
New business models
Several presenters introduced new business models for small cell deployment. These included fixed broadband operators such as Virgin and COLT who offer a managed service to install, connect and physically maintain small cells in urban areas. The access points would be remotely configured by the host network operator and use their spectrum, but allow them to scale up and deploy large numbers quickly. A managed service also avoids heavy CAPEX investment, making it attractive to finance.
Cloudberry, a new Norwegian startup led by several ex-Network Norway executives, also offers a managed service to European network operators seeking to deploy mainly residential and enterprise femtocells. Using their expertise, they will host a small cell gateway and handle all the logistics. Targetting especially the 3rd of 4th operator of each country, and possibly even some MVNOs, they offer their expertise and have partnered with NEC for the outsourced solution.
Avanti divulged detailed financial costs of their broadband satellite service, aiming to disprove the impression that satellite is an expensive option. It's particularly suitable for remote/rural areas using satellite terminals costing around £900 and a ruggedised femtocell which can be implemented by any satellite TV installer. This may be attractive to provide voice and some data coverage in otherwise unreachable areas at a commercially viable price point. I wondered if there was a business opportunity there for someone to target remote communities and capture occasional roaming traffic too.
Wi-Fi offload, seen and unseen
Telefonica O2 have recently invested in their own free Wi-Fi service – registration is required but it's open to all. One London district is covered with one area delivering the equivalent of 1,350Mbps Wi-Fi capacity per square kilometre, with an average of 15Mbps per access point and total network traffic of 400Gbytes/day. It offloads some 5-10% of data traffic from their own 3G network. Later phases involve upgrading these Wi-Fi hotspots to provide 3G using an integrated small cell.
Virgin Media, the largest UK cable network operator and MVNO, serves 4 million customers and offers broadband service of up to 100Mbps. They've also been providing free Wi-Fi throughout the London Underground this year, connecting a 1Gbps fibre backhaul to an average of 13 access points. With no interference underground, their Wi-Fi service delivers high speeds (typically 18Mbps) and services 600K registered users with up to 900K sessions per day.
The conference venue itself was a good example of how Wi-Fi is used today. While speakers discussed the fastest 4G data speeds and wide variety of options available in mobile networks today, there was no useful mobile phone coverage from at least 2 or 3 networks in the main conference room on the first floor. Many delegates took advantage of the free Wi-Fi service that required the usual clunky login through a web page. Over 2 days, more than 25GBytes was served, of which one delegate consumed 5Gbytes. I wonder if the main reason that mobile networks aren't seeing even higher traffic levels is because they simply aren't providing service in so many places where customers would actively use it, and aren't measuring/recording the vast amount of traffic bypassing the network over Wi-Fi. The remote control to advance the slides also shared the same 2.4GHz Wi-Fi spectrum and occasionally failed to operate. Whether that co-incided with heavy Wi-Fi congestion because delegates weren't listening to the presentation is left to the reader's imagination.
A large part of the operational cost of a mobile network is energy, and several speakers explained what they (and the industry overall) are doing about it. Speakers from Orange and Telenor brought the good news showing much lower energy costs per Gbyte. Even with the huge growth in data traffic carried, networks have overall shown only a slight increase in energy consumption of 1-2% worldwide. Further good news is that studies of small cell deployments show overall energy savings, bringing not just financial benefits but also reducing CO2 emissions for the planet.
Julius Robson explained the NGMN backhaul requirements document, which has been very well received by operators (especially in the US). There is no single best solution and operators will deploy a range of options from wireless (Non-Line-of-Sight, Micowave) to wired (Fibre, DSL, Cable). Taqua suggested that operators plan to use NLoS in anything from 20 to 80% of their public small cell deployments, although most would be nearer the lower figure. They've had successful trials with two Tier 1 North American operators, providing 60Mbps to each of 4 remote small cells (240Mbps total capacity). The sweet spot seems to be 100 to 700 metres range, with backhaul capacity designed to carry average rather than peak end-user datarates. Several vendors reported significant progress in backhaul products for small cells (including small/low cost point-to-point microwave) and I believe that there is lots going on to address the issue including much that is not yet public.
The general impression I came away with was that most European operators are planning their major radio capacity investments to follow a sequence of UMTS900, Wi-Fi offload, 3G small cells and then LTE. This is a really complex business and technical problem with many potential solutions and operators are actively grappling with which ones they should place their bets on.
Sitting in a conference room without any useful mobile phone coverage while being told that there wasn't a capacity problem on mobile networks seemed somewhat surreal to me. I'm glad I don't go to food catering conferences where I'd starve.
So I foresee that some of the new business models presented this week, combined with radical changes in the way in which operators plan/operate/maintain their networks will result is major disruption ahead. Today's hot topic of backhaul will make way for some of the logistics and scaling problems when substantial numbers of public access small cells are rolled out. This might lead to some of the new business models, vendor solutions and outsourcing arrangements discussed.
The outcome of this conference was by no means a consensus view on the way forward and I sensed a range of views from complacency to confusion. This suggests that much more information and results will need to be shared as the industry addresses the growing data capacity and coverage problems.
Apologies first for the longer length of the report and secondly the omission of many other points make by speakers during the event.
Below is a view of Andy Dunkin from Vodafone expounding their OneSON vision at the event: