Martin Wren-Hilton tells me he's never had so much fun. His job as Head of Mobile Innovation at TalkTalk is to figure out how to squeeze a high quality mobile network service from a slither of previously unused spectrum. He presented at MWC on what's involved, their test results to date and how it could evolve.
This UK company was spun out of Carphone Warehouse and has become a nationwide fixed network broadband supplier to over 4 million customers, using Local Loop Unbundling. It's also setup as a "thick" MVNO (with its own HLR, core network and real-time billing system etc.) using Telefonica's network. This service is initially only offered as an option to existing customers.
TalkTalk CEO, Dido Harding, has previously gone on record stating that they do have long term plans to deploy residential femtocells. They operate as a value end of the market, in a very similar way to Free France.
Guard Band Spectrum
In 2006, Ofcom auctioned off a 3.3MHz slither of FDD spectrum in the guard band between GSM and DECT at 1880MHz. Intended for low power use, it's most relevant for indoor use but there is no restriction against short range outdoor service. Few of the 12 licencees have made much use of it to date. TalkTalk won access to the spectrum for just £155,555 (approx. $250k) – not bad for full nationwide coverage.
Originally it was thought that it would be used for GSM only, and we reported an example with Tesco retail stores back in 2009.
TalkTalk setup a research project to establish whether the spectrum could be used for LTE, identify the technology platforms required and understand what form mass market deployment could take.
Designing a "Super-Router"
They built their own prototype femtocell based on the Broadcom chipset, Node-H software and Arcadyan hardware with a Huawei core network. The result is an "all-in-one" ADSL/VDSL modem, 802.11ac Wi-Fi router and LTE femtocell that supports high speed LTE voice and data. The RF power level for LTE is even lower than for Wi-Fi, improving battery life, voice quality and data speeds but with a wider coverage footprint.
Traffic management is important, and they've incorporated QoS for both uplink and downlink, Self-Organising Network capability and remote management using TR.069.
Performance and Compatibility
Lab testing has shown that achieving data rates up to 16 Mbps is quite feasible. Multiple concurrent voice calls are also supported, although it was not stated how many concurrent users the femtocell can support.
Although the frequency band and mode are defined in the 3GPP specifications, nobody else has been using this configuration. With some trepidation, Martin went out and bought 30 different models of LTE smartphones off the shelf and tested each one – they all worked perfectly.
Research from Qualcomm has proven that coverage extends up to 60m from the residence. The LTE signal is more effective than that of Wi-Fi and doesn't have to compete with other uses on the unlicenced spectrum bands, such as baby alarms etc. It's also at a lower frequency of 1.8GHz rather than 2.4 or 5GHz.
Qualcomm estimate that good street coverage can be achieved from a random distribution of 20 homes out of 320 houses, just 7% penetration. TalkTalk is already installed in about 20% of UK homes, so this means they should be able to cover many residential streets.
In the chart below, Red = best coverage, light blue = poorest but still adequate. The red squares are the homes with femtocells.
Martin played a brief video showing the comparative signal levels as he walked along the street outside his home. This showed the Wi-Fi signal dropping off quite quickly while the LTE signal remained very usable even at the far end of the street.
Co-ordinated Femtocells using X.2
Research has also included the X.2 interface used to communicate directly between LTE basestations and co-ordinate interference management. This isn't required with outdoor macrocells, which operate on Telefonica's frequencies, but is very useful between multiple TalkTalk femtocells in the same neighbourhood or apartment block.
Multi-vendor compatibility isn't an issue because the femtocells will be all under TalkTalk's control and share the same design.
Voice using VoLTE
In the UK, most voice calls are still carried over 3G although we can expect more rapid take-up of VoLTE over the next year or so. This femtocell would only support LTE voice calls, requiring support in the handset and the core networks.
With TalkTalk addressing the "value segment" of the market, I would say the viability of the project depends on the price of LTE handsets coming down.
Future expansion to support LTE-U
With both LTE and Wi-Fi in the same box, it should be quite feasible to aggregate these to achieve higher speeds. This could be done either using a simpler Link Aggregation or by using LTE in unlicenced bands. Once these are standardised in Release 13 and the industry decides which to adopt, the femtocell could be updated to support either or both.
Since this 1800MHz frequency is licenced, it should be compatible for use as the controlling LTE channel in LTE-U. Speeds of multiple 100Mbps should be feasible. This ties in with their plans for 1Gbps fibre to the home service in the long term.
TalkTalk clearly have the technology that would allow them to build an "inside out" mobile network using residential Femtocells in a similar way to Free France. This expands the scope of a "Thick MVNO" by capturing the majority of traffic when at or near the homes of any of their subscribers, radically reducing the costs paid to the host mobile network.
This is a very forward looking research project, adopting LTE and leaving the future path open for higher speeds using aggregation in unlicenced spectrum. The commercial timeframe hasn't been stated but will surely depend on factors such as availability and penetration of low cost LTE smartphones.
By keeping access to all TalkTalk Femtocells open to all their subscribers and avoiding a white-list, it means they can serve their customers not only in their homes but throughout their entire neighbourhoods. It will also make the service much more seamless for the end-user.
I think this technology and business model are definitely capable of being very disruptive, just as they have been in France. Definitely one to watch, and probably visible to other potential challengers elsewhere too.