5G needs verticals, so where are they?

NGMN's second big 5G conference was long on alignment and shorter on technical diversity. And although there was a large cake, there were very few "verticals" around to eat it.

The following report was published on Friday last week in TMN’s weekly newsletter. To join the growing band of those already receiving the newsletter, complete this simple sign up here.

The NGMN Alliance was in celebratory mood last week. It is 10 years since its official launch, and in fact rather longer than that since three German operator CEOs started scheming to form an operator alliance that could articulate their demands for the next generation of mobile network.

20161012_202842

This is a large chocolate birthday cake. Your intrepid correspondent self-served and consumed a real (not virtual) slice, with very little latency.

Of course, in those days, it was a rather unsatisfactory 3G world and the next generation that gave NGMN its name was to be 4G. Now 4G is the current generation and so naturally the nominally determinative NGMN has concerned itself for the past two years or so with 5G.

18 months ago, NGMN held its first public 5G-themed conference, and much of the material from the conference stage was overtly technical. There were detailed descriptions of things like full duplex radio, massive MIMO, extended carrier aggregation, information-centric networking, new coding schemes, new frame structures, new waveform candidates.

This time around – not so much. Instead there was repetition of the sort of top level messaging that comes around when everyone gets a bit more poker faced, when a few dollars chucked on at a white paper author becomes a few tens of thousands chucked at a trial, a few million at a pilot. Jonathan Borrill, Marketing Director and sage of Anritsu had another explanation that also made sense: much of the early technical selection has in fact been done: it’s going to be a variant of filtered OFDM, a flexible frame structure, coding is nearly done, MIMO will be not quite so massive after all, we know the likely spectrum bands. Therefore there is no place any more for the technical hail mary passes. 5G New Radio is getting locked down. And as if to illustrate that, the 3GPP RAN1 working group was meeting at the same time as NGMN – with Wednesday’s meeting apparently lasting until 2am on Thursday. The rubber is hitting the road.

So this time around our celebratory and celebrated operators mostly stuck to the script – we need to collaborate with each other and with “verticals”, we need to guard against standards fragmentation, we must make 5G a success for society and the fourth industrial revolution. Motherhood is to be welcomed. Apple pie, I feel it incumbent upon me to inform you, is a delicious dessert.

God Send Verticals

If in 2G GSM stood for God Send Mobiles, then 5G awaits the arrival of verticals to save us. “Blessed be the verticals” is incanted through the C-Suite. Verticals, your job, your destiny, is to turn up and redeem this industry from itself, to hand out the Quells and suppress the nausea caused by operators’ suspicion of their creeping irrelevance.

And one or two verticals did show in Frankfurt this week – notably the major German car companies BWM, Audi and also Continental. The industry has just formed a joint cellular-automotive industry alliance by the way, so not to have brought these German names to a conference in their own back yard would have been an achievement. But there was nobody, not one vertically-blessed ambassador from agriculture, health, energy, utilities, transport, smart City government or even public safety. Not one speaker or panellist from any of these core, key verticals upon whose willingness to self-configure service levels across our sliceable, automated, programmable, massively scaleable and dynamically adjustable networks we are pinning all our 5G hopes.

Their absence was glaring and of the two explanations for the no show both are revealing. Either they were invited and did not come. Or they were not invited.

What would happen if the webscale players formed an alliance – you could call it FAGA – and asked these verticals to come and discuss how FAGA’s networks, their analytics engines, their very customers, could possibly be of use to them. They would, I would hazard, turn up. What would happen if FAGA formed an alliance, decided their key strategy was to design a product that would attract and whole new class of customer, and then never asked the customers? They wouldn’t do it would they?

OPERATOR PRESENTATIONS:

voda-slideJohan Wybergh sounded a note of caution. The new chair of NGMN and the still relatively new Group CTO of Vodafone said that the industry must not over hype 5G. He pointed out that although mmWave offers very large frequency bands and high throughputs, it’s not good at going through walls into the places where people live. He said that although we want verticals to engage there’s no clear demand as yet beyond enhanced Mobile Broadband, and it will also entail a major architectural change. And he raised again the issue of net neutrality potentially undermining the concept of traffic differentiation via network slicing.

Thibault Kleiner of the EC’s Digital Single Market was adamant that Network Slicing would be specifically exempt from Net Neutrality rules. It seems Europe’s telcos are not yet convinced of this.

Tom Keathley, SVP AT&T, said that the operator has been testing at 15GHz and 38GHz and at 3.5GHz, and that it is possible to achieve double digit Gbps to a single user and single digit Gbps to multiple users below sub 6GHz in a lab environment. But there’s a great deal of learning to do before it is implementable in a scaled way. He also said questions remain about the fixed wireless use case (fibre replacement) of 5G that Verizon is pressing so hard for. “Whether you can reach 5 or 30 radio heads in the home makes a great deal of difference to the business case,” he said. AT&T would move in line with standards, and would see 5G as a mobile-first technology, with fixed wireless as a subset use case.

This contrasts with Verizon, which is committed to an accelerated 5G development, targetting the fixed wireless use case as a first service. Adam Koeppe, for Verizon, said that field trials had been positive, as the operator has tested at 28GHz using a 100MHz bandwidth, that has the potential to be aggregated up to 800MHz. With beam forming and tracking experiments ongoing, Koeppe seemed positive about the fized wireless use case. And he even said that there are certainly aspects that can be taken forward to full mobile 5g – in terms of modelling channel properties and propagation, even though of course beam forming and steering will be a lot more challenging in mmWave when you are tracking mobile users, and not delivering to a static CPE inside someone’s house.

Samsung said that path loss at 30GHz was not as bad as it had feared – pointing out that signal drop-off through a patch of trees is not linear. In other words, you don’t lose more signal the more trees there are, you get an initial drop off, but then that signal remains.

Korea Telecom (KT) has also been channel modelling at 28GHz, and has set itself a deadline of 2018 to have some limited 5G services – very high throughput plus some low-latency-related services – ready for the Winter Olympics. It would use this to go to a wider commercial deployment in 2019, it said. That would put it a year ahead of the ITU’s IMT-2020 process, for instance.

china-mobile-slide

Here’s China Mobile’s 5G timeline and spectrum candidates

On spectrum, Orange’s Alain Maloberti said that the industry needs 5G spectrum below 6GHz, including being able to use existing LTE  bands for 5G – and it could also have 700MHz as a coverage workhorse for large scale deployment. He thought that 3.4-3.6GHz(up to 3.8 in Europe) has the potential to be a pioneer 5G band. He also gives priority to a band that he thinks has the best propagation and potential for harmonisation – with 24.25-27.5GH carrying the highest potential in terms coverage, bandwidth and synergies with 28GHz that are being considered by other geographies (US/Asia). He added that there is also potential in 6-24GHz bands not being considered in ITU, such as 6-8.5, 10-10.6, 21.4-22GHz. This would involve regional level discussions to see if these are available for 5G.

NTT’s Takehiro Nakamura said that this regional approach might result in a “partial harmonisation” on spectrum globally. He also said that the industry needed to be more aggressive on spectrum thant he path laid out by waiting for WRC-19 clarification. NTT wants to be ready for its own Olypics in 2020, and has already proposed the use of 3.5, 4.5 and 28GHs bands. Those three bands are the 2020 candidates and NTT “cannot wait for WRC-19,” he added.

VENDORS HAVE THEIR SAY

Aside from the anchor level sponsors – Cisco, Ericsson, Huawei, Nokia, Samsung, ZTE – network vendors were kept to panels and the exhibition rooms.

ericsson-slide

Hard to see where that top layer standardisation is coming from, Ericsson said.

For Ericsson, Ulf Ewaldsson said that there is still some uncertainty in where the command and control layer will be standardised. It could happen in SA5 but that is intended only for telco nodes only and we aspire to much more in 5G, he said. Is it going to be the TM Forum setting that – “it is hard to see”. NGMN tried to set standards for OSSBSS and it is hard. The answer might lie in Open Source projects: OpenO from China Mobile is one of those answers, E-COMP is one of those answers.

Nokia’s Moiin Hossein ran through the company’s 5G architectiure thinking, which you can see in more detail here.

Huawei was demonstrating a 5G trial with Vodafone, using a Filtered-OFDM waveform with Sparse Coding Multiple Access (SCMA) and Polar Code. It was also showing live video of a MU-MIMO trial taking place in China, using C-Band spectrum.

T-Mobile launched xRAN, a sort of decomposed eNodeB architecture, that seeks to put base station software (from Radisys and Aircent) and run control plane software at an abstracted layer in the cloud. The idea is to be able to places edge capacity where it is needed, but run control architecture centrally, or from a nodal point looking at clusters of access points. It’s a sort of SDN for the RAN.

Anritsu was showing mmWave channel modelling at 77GHz for in-building, and also modelling RF distortion and propagation over fibre, alongside VPIPhotonics, an area that may be of increasing importance as C-RAN architectures are deployed.

Comments

1

Leave a Reply