with Hossein Moiin, Chief Technology Officer, Nokia Networks
Though LTE and LTE Advanced are still being rolled out worldwide, the wireless communications industry is already gearing up for 5G, envisioned as its next technology revolution. Many in the industry are debating exactly what features and benefits should be offered by this next generation of wireless. FierceWirelessTech Editor Tammy Parker recently checked in with Hossein Moiin, CTO of Nokia (NYSE:NOK) Networks, to discuss his vision for 5G. Following is an edited and condensed version of that conversation.
FierceWirelessTech: There are so many different opinions regarding what 5G should be. What considerations do you think should drive 5G?
Moiin: I was intimately involved in designing 4G. What I've noticed in 4G and maybe also to some extent even more so I would say in previous generations--2G, 3G and analog—was that we as a community of network designers designed those previous generations with a network-centric view. We addressed coverage, capacity, power, the number of base stations, what is needed in the core network. We went around it in a very network-centric design paradigm. If I have one vision for 5G, it is that we change that paradigm into a user-centric network design, so that the design addresses first and foremost what are the predominant use cases of 5G, takes those verticalized requirements and creates a harmonized platform that can satisfy as many of those requirements as possible.
Is it a new network? Maybe that's the wrong question. It will most likely have a new air interface, and I'm fairly confident that will be the case. But it will more importantly be used as a shorthand by the industry to combine a lot of advances in system design, maybe in the architecture, taking advantage of things like cloud, network functions virtualization and software-defined networking in the non-radio part combined with the radio itself. Soon it will be a shorthand that we will use, similar to what happened to LTE.
When we were designing LTE we started with not the radio part but with the core part and we called that SAE, System Architecture Evolution. You don't hear about that anymore because LTE became the shorthand, if you will, for all of it. I use 5G in that sense, as a shorthand for a lot of new requirements coming from many different directions, many different industries.
LTE/SAE was designed as primarily a data network but a data network for a very specific use case, which was to take the fixed broadband experience of the time--so we're talking the late 90s-early 2000s--and translate that into a mobile environment and give the users a similar or better experience than they have on the fixed line. So with mobile broadband, we had a very clear use case.
Mobile communication and mobile networks, largely LTE but also others, are now being used by many industries and for many diverse applications, because mobility has inherent advantages over fixed. But also the fact that they're wireless and have very broad coverage has also real advantages over competing technologies. So many industries are looking at 5G as the next evolution of the network that will satisfy their requirements for low latency, much larger capacity, etc. So we put all of that into the 5G bucket.
Then we talk about is this really a new network, and the answer is clearly yes. It will have a different architecture, it will be virtualized, it will be using a new interface and what it will do is efficiently combine existing and new networks to deliver specific services to specific users. So this is the user-centric aspect that I talked about.
It's not realistic to expect that this will be replacing LTE anytime soon. There are many new use cases that this will enable, and so it will have a life of its own.
FierceWirelessTech: You brought up the fact that there will be a new air interface with 5G. I know non-orthogonal multiple access (NOMA) and Generalized Frequency Division Multiplexing (GFDM) have been proposed. Is a particular air interface for 5G favored at this point?
Moiin: Let me be more precise. I see it is very likely, but not guaranteed, that we will see a new air interface, and this will be a standardized air interface. We really can make significant improvements. You mentioned NOMA is one such example, and I think there are alternative technologies, maybe Generalized FDM is another one. Also I would not count out improvements in OFDM, similar to what we use in LTE and Wi-Fi.
Having said all of that, yes, it is very likely that we will see a new air interface and it will be standardized, because of the dynamics of the ecosystem and the size of the ecosystem requires that. But it is too early for me to talk about the details. What I need in order to make a determination is simulation results initially and finding out what other colleagues in the industry think so that we can come to an appropriate standardization body such as 3GPP and standardize [the interface]. So it's too early to tell what it will be, but most likely it's a new air interface because there's just so much to take advantage of.
One of the things that will distinguish 5G from LTE is the possibility of centimeter or millimeter waves, and this is a totally new area for us. It is required in my opinion for the simple fact that we're running out of potential spectrum for allocation and we need a lot of spectrum. Carrying 10,000 times more capacity in the network is not going to be possible with lower bands so we need to go to the higher bands and this clearly requires some more research, more simulation, more modeling, and we're working with some of our very advanced customers on trying to model the channels… to see what combination [of spectrum bands] would give us the best possible network that will also be useful for many applications that require very little latency.
FierceWirelessTech: In addition to the topics you mentioned, what are other important technologies for the 5G basket?
Moiin: I think maybe more significant are two aspects of system architecture to really improve the agility of the network to respond to changes, so how resources get allocated. Even with LTE, we have fairly static resource allocation. We decide that there's a need for a base station in location X, for 4,000 meters, this is how much power we give it, this is what kind of base station is deployed. In the future, we are pursuing much more dynamic resource allocation so that these kinds of questions are answered automatically. And automation is another key deliverable of the 5G era. That is required if you're going to have massively dense networks.
We've got to make networks much more intelligent. To me that is also an area of promising research and development, to make sure that we can take advantage of advances in artificial intelligence in other fields so that we can have networks that are self optimizing and self organizing like we do with SON but that are much more advanced than what we have today.
So in short, three areas I would say are critical. One is the new radio interface and its integration into the rest of the network. Two would be the new system architecture, dynamic allocation of resources. People call it network function virtualization, but I want to maybe go further and do it for the radio, also separate the control plane from the data plane--the common name for that is software-defined networking. But I want to take that to a different level so I can scale my control and data plane differently, all in a virtualized context. And the third is management of the network. Those are really the three key areas where I say we need a lot of efforts into research and development to make sure the networks are ready for many new devices, some of which will not be controlled by humans but by other machines.
FierceWirelessTech: Do you think 2020 is a reasonable timeline for introduction of 5G networks?
Moiin: We started exploring this probably a year and a half ago. But in my view, by November of next year when we have the World Radiocommunication Conference, we should have some preliminary results so we can motivate the ITU and other intra-government bodies into making sure that we have the sufficient spectrum and the right spectrum for allocation.
In Europe we have established the 5G public-private partnership together with the European Commission as well as other interested parties, both commercial-- such as Alcatel Lucent, Ericsson--and universities and research institutes. So we're exploring at this point. I think the system design and pre-study items for standardization will probably take us toward the end of 2017, at which point standardization begins in earnest. It will probably be finalized by mid-2020, at which point I think 5G commercial systems will be available.
A couple of deadlines are being given by our more advanced customers so that maybe by 2018 we should have some usable prototypes. They may be pre-standard, but they should be usable so that it can be shown what the capabilities of the new technology are. That deadline is primarily driven by sporting events. Following that, there is in 2020 another large sporting event, and our most advanced customer would like to see commercial deployment at that point. [Editor's note: Moiin later acknowledged he was referring to the 2018 Winter Olympic Games in South Korea and 2020 Summer Olympic Games in Japan.]
So broad commercial availability and deployment I see sometime in 2022 to 2023-2025 in advanced markets and later in the next generation of markets.
FierceWirelessTech: Do you want to offer any closing thoughts regarding the path to 5G?
Moiin: To me, the most important thing about 5G today is to minimize the hype and focus on the real work. And the real work today is about requirements, use cases, and making sure we optimize the network not only for the network's sake and for the network vendor, such as myself, or the operator, but also, most importantly, for the users.