An early assumption that 5G would be led first by densification has been challenged in the past year or two by the emergence of Massive MIMO as a major element in many MNOs’ 5G strategies. Operators, particularly in Asia, are looking to use large antenna arrays, combined with beamforming, to add huge amounts of capacity to their existing macrocells, before they move on to deploying large numbers of small cells.
European MNOs have been less aggressive about Massive MIMO trials than some of their Asian counterparts, such as Softbank in Japan or China Mobile. But the UK is proving to be an exception. BT’s mobile arm, EE, was an early public exponent of macrosite enhancement as more important, in the first phase of 5G deployment, than small cell densification. And Telefonica O2 – which has been a significant deployer of small cells – has got the MIMO bug too, and is now working with Nokia on two trials of the technology in London.
These are taking place in two neighborhoods with high levels of mobile data traffic – Kings Cross and Marble Arch. O2 is using the 2.3 GHz spectrum, which it acquired in this year’s UK spectrum auction, for the pilots.
“We recognise that customers’ need for mobile data in London and other urban areas continues to grow at a rapid pace. This is why we are working with Nokia to trial Massive MIMO and to explore the opportunities to provide the increased capacity and denser coverage for our customers, in the areas they need it most,” said O2’s CTO, Brendan O’Reilly.
The pilot will enhance O2’s current network capacity in London and lay the foundations for 5G. “O2 will evaluate the trials with a view to potentially deploying the technology elsewhere in London and beyond to enhance connectivity,” the operator said.
Three UK, the smallest UK MNO, has also been investing in Massive MIMO technology, from Huawei, to reduce the need to seek additional cell sites for densification. And BT has been testing the technology in Bristol, among other places.
Meanwhile, BT has been talking about its unique advantage in its home market – of being the only fully converged fixed/mobile operator, thanks to its acquisition of EE. Last week, it demonstrated a 5G system for relaying video footage of sports events to broadcast production facilities, at London’s Wembley Stadium. This could be cheaper and more effective than the satellite technologies which are commonly used for live sports broadcasting now, it said.
In the demo, BT used a prototype device from Huawei, which provided connectivity between camera equipment and a 5G antenna high up in the stadium. From there, according to Matt Stagg, BT’s sports director of mobile strategy, the signal is backhauled over a fixed line to BT’s production facilities in Stratford.
Stagg said that remote production is BT’s best-defined 5G use case outside the consumer world, so far. It will not only support live events, but behind the scenes footage – since the Huawei device goes into a camera operator’s backpack, it can be used anywhere (for interviews on team buses, BT suggested). It can also support commercial services for other broadcasters – BT could offer slices of its network to different sports and news companies such as the BBC or CNN.
BT says it aims to launch 5G in 16 of the UK’s largest cities next year, including the four capitals of Belfast, Cardiff, Edinburgh and London.
Another development which may boost the UK’s 5G roll-out is coming from Openreach, BT’s increasingly arm’s-length wholesale wireline group. Openreach has been a major force in developing G.fast – the technology for delivering fiber-like speeds over existing copper – as an alternative to fiber in some scenarios. The wholesaler has announced plans to use G.fast to boost speeds on its copper networks for 5.7m homes and businesses by the early 2020s. It will extend G.fast to 81 new locations, including parts of London, to deliver speeds of up to 330Mbps over the coming two years. This will be in tandem with its fiber expansion, which is targeting fiber-to-the-home services reaching 3m homes by the end of 2020.
In Australia, the National Broadband Network (nbn) was initially mandated to deploy fiber only, but after embracing G.fast, it was able to activate some 12,000 homes a week while FTTH activations lagged at around 7,000 a week. One estate in particular, comprising around 25 homes with particularly poor broadband speeds of between 1Mbps and 2Mbps, was fitted with a micro-node in the street to bring in 25Mbps speeds. The cost per premises for the nbn in doing so was slashed from around A$15,000 per premises to A$5,000 per premises – and the 25Mbps speeds were prior to rolling out G.fast equipment.
But G.fast is also a potential technology to support 5G backhaul and fronthaul, where fiber is too expensive or complicated to deploy. Industry players are beginning to embrace G.mgfast, a successor to the current standards, which has been designed with small cell backhaul and virtualized RAN fronthaul in mind, as well as access use cases.
The new candidate standard replaces XG.fast (also known as fifth generation G.fast), which was designed to shift the 212 MHz baseband ceiling on copper twisted paid broadband and lift it into the space between 350MHz and 500MHz. Now the plan is to double the 212 MHz to 424 MHz first, and then double it once again to 848 MHz after that.
We have already seen trials of aggregate (combined speed in both directions) throughput approaching 2Gbps and this should take it up to five times this speed, but also add in full duplex mode – already achieved with DOCSIS in the labs. That allows the full bandwidth of 10Gbps in both directions, using echo cancellation.
The new technology was demonstrated by nbn, providing 8Gbps over 30 meters of twisted pair over copper and 5Gbps over 70 meters, and G.mgfast should soon be pushed to be significantly faster over similar lengths.
The challenge that will take well into 2019 to address is how to cancel crosstalk in both directions at both ends of such copper wire, especially when it is used with both VDSL and G.fast in the same bundle. Not only will this not be easy, but it will take considerable processing power at each end and chips and servers will have to be beefed up.
The core considerations are that 5G will need one millisecond latency and the entire broadband set up will need to offer multiple levels of quality of service protocols, and support them even on over a single line, to support 5G network slicing – guaranteeing a single service a certain amount of bandwidth, without front ending each connection with specialized processing logic.
The G.mgfast standard, when it emerges, is also expected to support point-to-multipoint delivery – also taking a leaf out of DOCSIS, MoCA or G.hn’s ’s books – so that a single line can support multiple locations specifically small cells all backhauled on a single line.
This is a very ambitious rewrite of the copper wire broadband standards, predicated on two factors –vectoring has freed up copper wire from its largest constraint, that it is an unprotected wire prone to crosstalk; and the spend on 5G is beckoning copper.