Towards some answers to the key 5G questions

Part II of Keith Dyer's look at how 5G looks to be coming together, and some answers to the questions he raised in Part I.

Something to love/hate Twitter for is its ability to send a stream of meaningful, inspirational quotes past your eyeline. I saw one this morning: “I’d rather live with a good question than a bad answer.” This was attributed to Aryeh Frimer, who the internet tells me is a chemist and also a Rabbi (although like most quotations it seems Frimer was adapting someone’s adaptation of the original). Well, risking Mr Frimer’s wrath at misappropriating his profundity, here – a little later than intended – are some (hopefully not too bad) answers to the article “Questions that we must answer on 5G.

I wrote a piece last week that described some key questions facing those who would like to understand the journey from here to 5G’s introduction. Here, I attempt some answers, with reference to what is being proposed by some in the industry.

QUESTION: Why would a very low power, narrow band IoT-dedicated network need to be defined within the same “G” as something like a highly mobile, wideband nomadic node? Can it be part of the same G?

This question really asks – can a network be designed that can meet widely competing demands. It implicitly includes the question – can it be done through the same air interface, or at least through a common standard?

If there is going to be a split between “5G” networks, then the IoT is likely to cause and expose it . This is not just because of the disparities between performance requirements of mobile broadband and IoT, but between different IoT use cases themselves. NGMN has already identified ultra-reliability as a use case for certain critical functions (examples would be vehicle to vehicle, or critical/emergency communications). Another required function could be inherent security – both physical security and of the data itself. Yet there are other IoT use cases that carry none of these extreme requirements, and could quite happily toddle along on an as and when basis in terms of connectivity.

IoT 5G variation requirements

5G and IoT requirements Credit: Qualcomm

The advent and apparent success of low power, dedicated IoT network technology, plus the inherent capabilities of LTE-M and category zero devices may, in any case, take many “normal” IoT functions a long way into the future. What does that leave for 5G – presumably the more extreme use cases such as ultra-high densities, and the highly secure, available and reliable type of uses.

The air interface
What about a single air interface to meet competing demands?

The first thing to get clear here is a divide between techniques or capabilities that could equally be applied to the current OFDM-based networks, and the definition of a new waveform to support new use cases.

Examples of the former might be carrier aggregation, which in a 5G scenario might be adpated  to co-opt large bandwidths at higher frequencies. Or the use of more antennas with large scale MIMO. Or better interference cancellation or co-ordination technologies to enable things such as multi-RAT support and dual (macro-micro) connectivity. Or even something new like full duplex which may double available bandwidth by simultaneously sending and receiving in the same channels. Most of these could be applied in OFDM (LTE) based networks. They could also be adapted within an overall 5G scheme to reach the sorts of throughputs discussed.

This slide from FHH Insitute gives a good run down of what can happen in the air interface. Note that the waveform and new access coding is only a part of it.

5G Air interface

Credit: FFH Institute

But OFDM, due to inherent physical restrictions to do with frame size and its frequency guards, cannot meet some of those advanced IoT short-packet size, extremely low latency requirements.

New air interface for 5G

Why a new air interface? Credit: Alcatel-Lucent

However – to support these diverse requirements within a single air interface – many think we are indeed looking at a potential new waveform and coding schemes. Alcatel-Lucent is looking at Filtered-OFDM or UF-OFDM (which is the same thing as UFMC just by another name) . Huawei also thinks F-OFDM has potential. Samsung has also been further researching a waveform that has been around for a long time known as called FBMC.

Air interface Huawei

Huawei’s Air interface: from left, Duplex mode, channel coding, waveform, multiple access, spatial multiplexing

Proponents of F-OFDM/UFMC think its characteristics will enable it to perform better than FBMC, effectively being a more efficient waveform when bursts are very short. Where bursts are long (streaming) then there’s not much difference between the two in terms of efficiency.

There are still other contenders. But the important point here is that at the moment researchers think they can meet these competing demands through a single air interface although the air interface itself offers a great variety or even flexibility of features.

Converged service control
Another aspect to this proposal to meet these competing demands within a single G is to use advanced control technologies, driven by a new service architecture, to create policy-driven network slices applicable to the relevant use case and application. This is one way to solve the many use cases/one network conundrum. You in effect create many virtual networks (slices) within the one “5G” network.

control plane HuaweiFor example, this slide from Huawei advocates that combined control plane, with dynamic resource flexibility enabled by NFV, and control of the network via SDN.

You are essentially here looking at a network which looks at a use case, or user, and defines the required parameters – latency, throughput, level of security – according to policies. The “service” is then hooked up across the network, or it could even be deployed at a different point in the network (ie an edge cloud platform) if that is deemed to be the most efficient way to do it. Check out this slide from Alcatel-Lucent for a view of  a policy-defined network with APIs to deliver services in this way.

Credit: Alcatel-Lucent

Credit: Alcatel-Lucent

What is required here? Elements as identified by NGMN in its technology building blocks include NFV (including the V-RAN and cloud edge), SDN and inherent (not tunnelled) security. Let’s add to this the requirement for new optimisation and “cognition” in the network, driven by advanced data analysis. Are these technologies 5G? I think if you design them to support applications and use cases that you could not support in LTE or LTE-A then the answer is yes.

So yes, in that regard, when you add together the network and service control, the air interface and evolving access technologies, it’s possible to see 5G forming as a recognisable entity .

QUESTION: Can you do both? Design networks to specific use cases and also have flexible enough technology to meet as yet unimagined ones?

This is a different sort of question, and it comes back to something I first heard proposed by EE’s Network Architect Andy Sutton and reflected in an Economist piece that everyone got excited about last week (nothing like a bit of mainstream validation I suppose). And that is that if 5G is done right, it could and perhaps should be the last G. That is because it will provide a flexible enough, horizontally capable, platform upon which incremental technology can be deployed as needed. This contrasts with what we understand so far – the vertically integrated per-service network.

Information Centric Networking 5G

Information Centric Network. Credit: Cisco

And yet… there are those who highlight that there are radical post-IP, post packet-networking technolgies coming along. These are things like name-based networks, as proposed within Information Centric Networking. Paul Makiewich, CTO of Service Provider Mobility at Cisco, was strongly of the opinion that as 5G goes forward, network architects should also be aware of momentum in networking in the wider sense.

Is 2020 enough time to take this into account? No. So 5G will, from a network architecture point of view, therefore not be the last G.

Can you therefore design to specific use cases and unimagined ones? Philosphically, yes – but only if you think you are building something that will effectively never need to be rebuilt. Practically?I think it’s only fair to conclude that open-ended future flexibility is not a requirement, it’s a dream .

QUESTION: Who has to unpick all that into something that is 5G? Can NGMN present its requirements and also somehow make sure these are processed within the standards organisations as deliverable standards that will, as deployed technology, deliver on those requirements?

The simple answer here is that everyone works towards IMT2020, which is yet to be defined. The ITU’s ITU-R is set to produce its vision for IMT2020 this year,  setting the agenda for the World Radiocommunication Conference 2015.

Standards bodies such as 3GPP, IEEE will be working to the ITU-R working group within that. Different players can define their requirements, then, into those bodies. NGMN as a body of operators has some heft – NGMN exec and Deutsche Telekom CTO, for instance, told TMN that he was confident that NGMN’s requirements can be reflected by the SDOs, given that representatives from many of the same companies sit on many of the relevant bodies. But it has to be said, NGMN is no more defining 5G than any other party – it remains a group of operators proposing its ideas. That it claims to have been mostly responsible for the success of 4G (not everyone agrees with this view of history, by the way) is also no guarantee of it achieving the same 5G relevance. Requirements are not standards. Saying you will start with requirements and then move to technology and then standards doesn’t mean it will happen that way. The vendors in particular are likely to take a more technology-centric view, although of course they are also very aware they must develop to a potential market, rather than trying to create a market.

QUESTION: Are those involved mature enough to admit that inconsistencies and contradictions exist and to do something about that?

I said that due to the process we are seeing a lot of technology being thrown at 5G, and a lot of claims for what 5G is for. When that happens it can create a situation where you can easily ask: “How can X and Y both be a part of 5G, where X and Y cannot co-exist”. Recognising those possible contradictions requires a maturity and understanding of where the other party is coming from. But nor is this a “gotcha” game – it doesn’t follow that just because you can identify that there is an inconsistency between different requirement, or that a proposed technology doesn’t meet those requirements, that “5G”as a concept is invalid. On the other hand, companies that are knowingly conflating existing technology directions with 5G are not helping either, as it exposes the technology as not “5G” and casts shade on what 5G is.

As I said in Part I, 5G is a destination that will be defined by the journey.