Vendors are generally delighted to raise their profile by pretending to have "5G" systems just around the corner, but it was a CTO from Alcatel-Lucent who threw a welcome bucket of cold water over the 5G hype at last week’s CTIA Super Mobility Week conference in Las Vegas.
Michael Peeters, the company’s wireless CTO, told the conference’s Towers & Small Cell Summit stream that the current talk about 5G development was “ridiculous” because it was focused on the old issues of technologies to increase capacity and bandwidth, whereas the real revolution would come from an architecture which was open, federated and “invisible”.
He made a very good point – vendors and operators are largely staying within their comfort zones, thinking of 5G as an extension to 4G, with or without a brand new air interface, and focusing on extending current technologies, such as MIMO antenna arrays, rather than on a completely new approach.
Yet with many physical network technologies approaching their limits, the real challenge is to enable a host of different platforms to work together as a seamless whole, largely software-controlled and flexible enough to support any usage pattern from low bandwidth smart meters to future generations of super-HD video.
While this big picture gets discussed, many of the actual R&D pro-grams and so-called ‘5G’ demonstrations are firmly centred on the same assumptions and building blocks as the LTE standards, with scarcely a thought for virtualisation or real heterogeneous networking (with scores of connectivity and spectrum options integrated, not just cellular and Wi-Fi). As Peeters put it, companies are throwing in all the elements which have not yet been included in official 3GPP standards, such as massive MIMO and millimetre wave, and claiming these innovations and associated patents should be considered 5G – well before any standards have been defined.
Don’t forget LTE and ‘4.5G’
Peeters’ own view of 5G is that it will create a fully open frame-work which will be invisible and “non-disruptive” to users and operators, enabling huge numbers of services and vast capacity be-hind the scenes through its flexibility, but leaving providers to differentiate with their applications and services. This platform will evolve after an interim “4.5G” stage, which will be more conventionally focused on extending current architectures to support greater capacity and cost efficiency. In particular – toeing the ALU company line here – 4.5G will involve very dense networks of small cells, to slash cost of ownership of networks in congested areas.
That is a pragmatic message echoed by the 4G Americas trade association, whose president Chris Pearson pointed out this week that, despite all the 5G hype, there is still a lot of mileage untapped in LTE. Indeed, of the world’s 321 LTE networks, only 20 are yet supporting LTE-Advanced capabilities such as carrier aggregation. "Most of our operators are fairly reserved in stating what they're going to do and when because there is no definition of 5G," Pearson said in an interview at the CTIA event.
And Kris Rinne, outgoing chair of 4GAmericas as well as an SVP at AT&T, added: “We want to make sure that we do not lose sight of the value and the runway that is left in an LTE roadmap as we continue to explore migrating to other capabilities and technologies.”
She believes 5G will be composed of "a network of networks” combining today’s and tomorrow’s technologies, but “not a new radio access technology deployed ubiquitously across the country.”
While most expect virtualisation to play a big role in the next generation of networks, Rinne wants to step away from the 5G label here too, noting that many virtualisation functions will not be attached to standards and will be implemented well ahead of "5G".
Indeed, technologies like virtualisation and densification could enable most operators to defer investment in "5G", whatever that may be, until well beyond the 2019-2020 timeframe set down by some early movers like NTT Docomo.
Instead, carriers could wait until 2023 or later to make a major architectural shift, and even then, it will not be of the kind they made to get to 3G and 4G, but will be all about the cloud and software defined networking (SDN). That need not be a dramatic change either, since operators are starting to lay the groundwork for a software-driven network already. Operators are already moving to “federate everything”, and not just through activities like NFV.
Peeters sees carrier aggregation as a way to federate spectrum, while another example of federation within the access network is the ability for a device to connect simultaneously to a macrocell and a small cell. Combining these techniques with virtualisation could reduce the total cost of ownership of a RAN by over 40% while improving user experience, he said.
Millimetre wave projects:
For all his wise words, vendors and academic institutions will continue to invest in their comfort technologies, and of course, however virtualised everything becomes in the coming decade, it will still be vital to keep enhancing the capacity and performance of the hardware elements that cannot be turned into virtual machines – back-haul links, antennas, radios and so on.
They may increasingly be managed and balanced from the cloud, but they still need to do their physical work as efficiently as possible, and they will need more and more spectrum to do so.
That has put MIMO and millimetre wave bands at the heart of current pseudo-5G initiatives. One of the projects focused on the high frequency, high capacity bands above 10GHz is based at NYU Polytechnic, whose wireless research centre is looking at using spectrum all the way up to 300GHz for "5G" networks in dense urban areas. The team has demonstrated end-to-end transmission of an HD video stream in the 60GHz range, with the prototype including modules for synchronization, equalization, and turbo decoding.
“By 2020 this is going to be in full-fledged development if not deployment. We’re not that far away. A lot of activities are going on today doing early prototypes,” NYU Wireless director Theodore Rappaport told EE Times. “I expect standardisation [of 5G and millimetre wave systems] to start in the next year, and you’ll see 5G being woven in, sort of dovetailed in, a graceful way to extend today's 4G LTE standard.”
The traditional downside of millimetre wave frequencies is their very short range, which is why they have come into the limelight along with small cells. However, in NYU Wireless’s tests in Brooklyn, advances in antennas and beamforming extended the signal’s range to over 1,000 feet or several blocks, without the need for line of sight. “You can be several streets away and around the corner, but energy stays in play. There’s a lot of multipath, a lot of reflection, and surprisingly a lot of scattering,” says Rappaport.
The project used beam combining - an omnidirectional antenna with a directional one – to achieve these results, with the directional unit overcoming some of the challenges of penetrating walls and foliage at high frequency. With current prototypes, operators would need 64 cells per square mile in a dense area like downtown New York City, and that architecture would support speeds of 2Gbps to 6Gbps per cell, though Rappaport envisages those rates rising to “tens of Gbps”.