Develop a Solid Understanding of 5G
There are quite a few important items we need to walk through. The first one is on the key performance indicators, I want you to understand what 5G is expected to do in the future. We then talk about some technology enablers. So, I want you to get a feeling of what is really novel in 5G, what are all these exciting things the engineers talk about. And finally, I’ll introduce you into the world of standards, why are they important, and how do they operate, within the setting of 5G? Let’s get going.I would love to be, that you understand the key performance indicators, the KPIs in 5G, appreciate the use-cases, have a good idea of the tech that’s being used today, as well as, have a good understanding of the 5G standards and procedures.
5G Network Key Performance Indicators
Now let’s come to the key performance indicators, and I always say 5G is both very boring as well as very exciting. It is boring because it follows the trends we have observed over the last years as we move from 1G to 2G, 3G, and 4G. Things always improved by an order of magnitude, data rate always multiplied by 10 or 100, number of devices increased, criticality improved. Now, the exciting thing about 5G is that for the first time it is so good that it can support a lot of the industry processes. So latency is very low. Hey, it’s so low, I can replace my physical cable, my ethernet, with a wireless system now. And this is what is one of the most exciting developments within 5G. Let’s compare the 4G and 5G KPIs.
So you see the table here, which 5G PPP produced, and I want to highlight three of these KPIs. So, the user experienced data rate, so we’re not talking about peak rate, that’s what you’ll experience in average. So in 4G it’s about 10 megabits per second, and in 5G we want your experience to be 100 megabits per second, right? This is stunning, imagine on average, you get 100 megabits per second. The user plane latency, that means the latency you get when you start opening a website or you start controlling things remotely. In 4G we can do 10 milliseconds and in 5G we want to go down to 1 milliseconds. And I’ll tell you later how this is being done. So that is a huge challenge on how do you really diminish the latency? In terms of the connection density, in 4G we can do a lot, 1,000 devices per square kilometer. Arguably we can probably even do more. And in 5G we want this to be millions of devices per square kilometer, so really upping the game by several orders of magnitude. And these are the 5G technical use-cases we are thinking of. Think of it like a pyramid, which is shown here. On top is what we call the Enhanced Mobile Broadband, it basically means more data to you as a consumer.
If you before took maybe a minute to download your movie, well hello, in 5G you can do that in a few seconds. So, we see that as a very, very strong technical use case, because of course we have more and more data hungry applications coming out. But we also have an IoT application here, which is on the bottom-left corner, which we refer to as the Massive Machine Type Communications, where we have loads of IoT devices, low cost, low on battery life, covering huge areas. And then we have the more exciting one and the really difficult one to do from a tech point of view, it’s the Ultra-Reliable and Low Latency Communications, or the URLLC, no idea who came up with this abbreviation, and that is really about mission-critical stuff. That is new, that is think of an ethernet cable being done wirelessly. So we are really trying to disrupt here heavy industries which normally would rely on a totally different technology.
5G Network Technical Enablers
Let’s look at the spectrum. And 5G offers a very rich and high plethora of spectrum bands, and roughly we have three bands which are new, and we call them the pioneering bands. There’s one which is below one gigahertz, so this is typically around 700 megahertz, in the UK, that’s the TV whitespace band. It has great propagation characteristics. So you put up one single base station, it has a wonderful coverage, well, yes, because it comes from the TV bands, right, so TV needs to have a huge coverage in urban, suburban and rural environments, so that’s really great for coverage. Great coverage means, you know, it’s good for investment.
You don’t need to put up so many base stations. On the other hand, of course, you get a lot of interference from other cells, because everybody can hear each other over very, very long distances. And there’s not too much spectrum available, so data rate is never too good. But it’s a great band to control things, all the real-time voice traffic can go in there, and some of the video traffic as well. And then, we have the 3.5 gigahertz range, and in each country, it’s a little bit different, but you know, that’s a good band and a very prime location, because most cellular systems today work on a band which is very close to that frequency, which means that all our planning exercises we have done, all the deployments we have done, we can reuse that for this frequency band. And the technology you want to use there is massive MIMO, loads of antennae raised, and I’ll explain that later. And then we have the millimeter wave bands. Now that is very new, that’s pioneering, and it is something everybody is very excited about, because we are using very high frequencies to support a very mobile scenario with loads of handovers and mobility.
Nobody was sure if that would work, really, but we are seeing now that actually, the propagation characteristics are so much better than we thought. There isn’t too much blockage, you know, there is a relatively good link, and the designs have been done really well, so we’re excited about having these millimeter wave bands where you have a lot of bandwidth so you can pump through tens of gigabits per second of data, downloading an entire Netflix movie within milliseconds. So these are the three bands I want you to remember: Sub-gigahertz, typical cellular 3.5 gig, and the exciting new bands of millimeter wave 26 in Europe, and 28 in the United States, rest of the world. So what type of key technologies out there, you know there are totally new ways of deploying your stuff in the field, and this is highlighted in red, here. We’re talking about, you know, device-to-device. Hey, you know, suddenly in 5G, you can start talking from one device to another. I have to say though, that in 4G, we could also do that, but we believe in 5G, that becomes really operational. Then we have very, very dense networks, so we are able now to get a density which we have never seen before. And we can have so-called Cloud-RANs, so rather than having the processing done on every single base stations, which you would sometimes see on the roof, you have only antennas in there, and these antennas are connected with a fiber down in the basement where you would have a server farm, which you have seen on the video before, and that server farm would process essentially all the data you need to process, and do all the processing. That’s called a Cloud-RAN. So there are totally new ways of doing that.
We have new physical air technologies. So massive MIMO, I alluded to that earlier. So this is about having loads of antenna raised in a single base station as well as in the mobile terminal, because we figured out that as entotically as you go up with the number of antenna elements, you can get rid of interference. So we’re talking about initially hundreds of antennas, and maybe, one day, thousands of antennas, who knows, you know, technology moves so quickly. And then we have millimeter wave, which works in the high frequency ranges, when we have a lot of bandwidth available. But I think honorably the most exciting thing is actually what is highlighted in green, and that is a totally new way of running your infrastructure.
So rather than being a business where we have boxes and cables, you know, 5G will be a business which is all about software. And that means that if you want to implement a new feature, you don’t need to send your engineer to do the changes and change the box there doing your cabling, you do that with a mouse click from your coffee table, right? And we are able to do that, because we decoupled hardware and software. And that is what the network function virtualization achieves. We decoupled the control plane from the user plane. That is what SDN does. And that allows us to do network slicing, which will I explain further down the road. So, I want you to remember these totally new ways of essentially using your deployment, new technologies on the wireless interface, but most excitingly, a totally new way of deploying your infrastructure and have a very software-centric view. Let’s deep dive into this softwarization.
So I say, that’s the biggest revolution happening. So if you look at this slide here, on the lower part on the bottom of this slide, you see my actual physical infrastructure. So we’re talking about computing hardware, storage hardware, and networking hardware. And you put a hypervisor on top, a virtualization layer, an operating system, whatever you want to call that, and that virtualizes my physical resources. So I can now suddenly offer virtual computing, virtual storage, and virtual network. You know, it’s being done on the computer you’re watching this article from. The operating system is virtualizing the hardware infrastructure so you can have many applications run in parallel. And that’s what we want to do, right? And once you have it virtualized, you then can have so-called virtual network functions use whatever they need, right? So one of them may need 10 percent of the computing power, 50 percent of the storage, and 80 percent of the network.
Hello, my hypervisor can offer that to the first VNF in the game. And this VNF might be, it can be anything, really. It could be my entire coordinate work, it could be just a security authentication function, it could be a handover mobility function, so think of VNF literally packages, which have an input and output, and something happens in between, but it’s virtualized. So it can use any resource it wants, and it can be ramped down, it can be ramped up, it can be moved in the infrastructure, because it’s a piece of software. Of course, you need to manage that, and you have the VNF managers, and the network function virtualization infrastructure managers on the site.
5G Network Standardization
Why is that important? Why is softwareization important? Well hello, because 5G is not a single, homogeneous technology. It is a whole set. A salad of technologies. Just look at this slide here, so many technologies out there. You got totally new ways of handling IUT: Cloud RAN, mmWave, Ultra Dense Networks, Vehicular Two infrastructure project, et cetera, et cetera, et cetera. How do you manage that? There’s no way you can actually deterministically program and configure that. Doing that in software, to control everything in software and manage it is the only way forward.
Who is who in the 5G standards world? You see two families here. On the top you see the actual standards initiatives. Something really global here. And you see 3GPP which is really like the operating arm of all the tel-co industry. This is where we standardize all the tel-co equipment. That’s the body which is responsible for you being able to make a phone call whether you’re in Sydney, or you’re in Los Angeles, or you’re in London or Beijing, wherever you are. So that’s 3GPP. The GSMA is an operator-led initiative and it came out of GSM2G in that they’re pretty good at actually getting the requirements for the system up and running. And so is NGMN.
So NGMN is a purely operator-driven initiative where they’ve set out requirements on what really drives essentially, often, the 3GPP and the ITU type of standards development. The ITU is an SDO, standards defining organization, and they are setting out, really what is 4G, what is 5G, what 6G will look like. And that’s the role of the ITU. And the IEEE; we are familiar with as well is another SDO which is taking care of standardizing anything which is often in the layer one and layer two technologies. And then you have loads of associations, call them lobbying buddies, all over the world. In Europe, we have the 5G PPP, and you see all the other names here on the slide. I would like us to understand 3GPP a little bit better. So let’s do a deep dive on how 3GPP functions and operates. So you have three stages really, or three groups in general.
So you have a Project Coordination Group, which is a very high-level group, kind of the directors of the game. They decide what should be looked at. Then you have the Technical Specification Group which defines, then, out of these high-level items specific work and study items. It looks at what’s possible, and then it approves certain technical decisions. And that is being then put down into the Working Groups who actually do the hard work of defining the actual standards, the standards, how the protocol works, how the architecture works, et cetera. And once that has been finalized, goes back to the TSG and they are then approving that. That goes back to the PCG who are then giving it the rubber stamp on that and it becomes a standard. And these are the structures. That’s the 3GPP operating structure. We used to have four columns, and now we have only three. So we have the TSG RAN, which is the Radio Access Network. And you see the different working groups here. So RAN1 is Radio Layer. RAN2 is Radio Layer two and three.
Then we have RAN3 taking care of the different interfaces, et cetera, et cetera. Then we have the TSG SA, which is the Service and System Aspects. So, again, you have the different groups there. You look at, for instance, SA2, which is the architecture group, very important. SA3 Security. We also have Codex, et cetera, et cetera. And then we have TSG CT, which stands for Core Networks and Terminals. And, again, very important work related to the core network design and the terminal design as well as the testing environments. Let’s look at the timelines, the 5G timelines. So it turns out that we have completed Phase One already, and that happened in the summer of 2018. We released 15, and that was the first step towards 5G. But, interestingly, it is not yet 5G compliant as per the ITU requirements. So we still need to do a lot of work and that’s being done. And in December 2019, we hopefully will freeze then the next release which will then be submitted to the ITU for approval, and if the ITU approves that, then we will have, finally, a proper 5G system up and running. – Now let’s do A Little Bit of Gossip to finish this off. You won’t hear that too often from anybody.
Let me talk a little bit about 3G. So people think 3G has been really a very difficult process to push for, in reality it was Qualcomm, an American company which came to the table of the standards and said look this is a working prototype. And everybody thought wow this actually works! Let’s standardize that! And the rest, as they say, is history. Was it the better system back then? We don’t know. Well actually today we know there were other systems out there which in theory should have been better. What else can I talk to you about, WiFi! You know I worked for a big operator for a while and I can tell you that WiFi was something the operators did not like for a very long time, and a lot of the engineers sometimes went in to look on what WiFi standard was doing, and maybe even slowing it down. Shh, don’t tell anybody.
Now the other thing is all about 4G. A lot of times you will see on your mobile phone hey there’s 4G I have good connectivity. In reality, by standards understanding, you’ll not always have a 4G system. 4G is if you really get a certain carrier aggregated bandwidth, which is not always the case. So it was more of a marketing gimmick to start with. I believe, as of today, almost everybody has a 4G system, but you know it wasn’t always the case. And a final Little Bit of Gossip about narrow band IUT. That is an exciting technology to connect all these sensors on the planet. And that was a technology which wasn’t planned to be done until 2021, ’22. And suddenly new technology emerged like Savox and Lara and everybody is like, hey we don’t want to lose that market opportunity. So the big boys in the industry started to design a connectivity technology which would allow us to connect all sensors in the world. And that is today’s narrow band IUT. So A Little Bit of Gossip from behind the scene. If you want to know more, just get in touch with me!
Reviewing 5G Technologies
We’re going to be talking about spectrum. Well, yes, wireless is all about spectrum, so you need to understand at least the high-level understanding what all that spectrum is about in 5G. We then move on to novel, exciting connectivity technologies which are being used within the 5G ecosystem. There will be some really mind-boggling technology I’ll be talking about, something nobody has ever really used before in a wireless setting. We then move on to more of the architecture side, the access network. Hey, how do you connect all these base stations on the roof? We then move on on the core network, the really transformative way how 5G will be very different to any other G you have ever had before. Let’s get started. So what are the learning so far? Well, I would like you to get a very solid understanding of the 5G technologies, the spectrum. You know, what type of interface we’re using, as well as the access and core network, and finally, we will spend a bit of time on this authorization of the 5G architecture, so I would like you to understand in great details, what is SDN, NFV, and slicing ?
5G Spectrum
I would like us to start by recapping the different bands used in 5G and what are the trade-offs. So just looking at this here,
you can see we have two rough regions. One is the sub-six gigahertz, which is 3.5 gig and below gig what we talked about. And then we do have the millimeter wave. Anything below six gigahertz is very congested. It’s prime spectrum because hardware’s cheap, the actual deployment is quite cheap. But the problem is it’s very congested. It’s very difficult to find a free band of significant bandwidth, which is available around the world. So we’re starting to move into the millimeter wave range where you see so many more free bands available today. So what is the challenge in 5G? The challenge is to bring together so many philosophies, so many different bands, so many different ways of doing things when it comes to spectrum. So just let’s have a look at the different bands.
Here are three bands I talked about. The methods today, we have licensed: fully licensed, shared license, we have end-licensed regimes. So whatever license you need you can do what you want as long as you obey a certain performance barometers. And the license shared by approach, you need to share with other companies in the area. So there are different ways of doing that. And then we have different aggregation methodologies. We can do a carrier aggregation per link. We can mix technologies here. Well, you know in 5G that all needs to come under a single hat, because we want to support a very, very wide variety of technologies and licensing methodologies. So if you look at it from a band and licensing trade-off point of view, let’s look for instance, first off, the applicability with respect to the band.
So let’s come back to the use cases. Remember, we had the mobile broadband, the EMBB, the IUT use case, which is the MMTC, and the low latency critical comms, which is the URL LLC. You know the sub-gig is actually pretty good for most of these use cases, and the 3.5 gig is also very good for the use cases. But you know, anything above that, we start to see that, particularly to support anything related to machines and other things becomes very tricky. Now when comes the application with respect to the licensing you can see from the table that unlicensed spectrum is not suitable for the critical communications, simply because you have no control over this spectrum. Anybody could do on the spectrum what they wanted. So you can’t really guarantee quality of service, quasis, we call that, and therefore it is not suitable. You need to use a licensed spectrum to do that.
New 5G technologies
So let’s look at one specific new technology, and that is mmWave. So mmWave is rated in the higher frequency ranges, anything theoretically above 30 gigahertz, we’re anything around 26 and 28 would still qualify as a mmWave. Now one of the downsides is that Pathloss increases, so if you have this same transmissions power to compare with a mmWave and a non mmWave transmitter, the range of the mmWave transmitter is much less. Then you have of course that problem that there’s a huge blockage effect as you go up in frequencies.
So, you know, certain radio frequencies stop being able to penetrate walls and windows and they get rather reflectant. Then we have the general atmospheric absorption which leads to the higher pathloss. We also have more noise because we have typically in the mmWave, a much broader bandwidth. More bandwidth means I’m capturing so much more noise, and then because of the very highly reflective nature in some of the difractions, we will have quite a bit of multipath visible in the mmWave channel. But you know what, let’s turn that from a challenge into a benefit.
So propagation effects not very favorable, so we will have a problem in terms of the network architecture because you know, we need to have much more cells to support the mmWave system. But at the same time, it’s a great benefit. We are increasing the density of my cells and therefore giving so much more capacity to the end users. Now let’s look at the signal characteristics. A short wavelengths, so it means good things, as is we can have small antennas, much smaller antennas, in fact, they can be so small that we can have MIMO movable antennas on your mobile terminal in the future. Now we have the problem with the propagation in general. You can have a lot of security features built on top of that, because to eavesdrop the channel becomes very difficult because we typically have established a very strong line of sight type of capability between the transmitter and the receiver.
So overall, huge challenges in the mmWave design, but that can be translated to fantastic benefits in the system. Let’s go now to another technology which is Massive MIMO. Massive MIMO is all about putting loads of antennas into my base station and also in the terminals. So if you look at the regular MIMO rate, and you would find it if you look on the roofs of the building adjacent to you, most likely. It is quite a long type of antenna. You typically have three cross polarized antennas in there so six antenna elements in total. And we do beamforming in 4G, it’s all good. But do you know Massive MIMO, we are really upping the game here, the massive MIMO rate I have on my roof is 8 by 8, so 64 elements, and each one of them is cross polarized, so 128 rf channels going out. Wow, that’s an increase from the six we have typically. And that allows us to have very narrow beams, we call them pencil beams, and not only in asamauph but also in elevation, we can do now 3D beam forming in a sense.
Right, so that’s quite exciting. And because you have so many more channels, you essentially get a channel hardening, so the receive perceive the channel almost ethernet like, so you know it’s almost totally constant signal once you start combining the signals from the different arrows. And then we can have a lot of interesting multi user cognizability, and have more spacial streams and also we need to trade one with the other, so loads of opportunities once you start introducing massive MIMO. But it all came from, you know, some interesting observation in the early days from Tom Marzetta and Bell Labs, and he observed that if you put the number of antennas asymptotically and your system becomes then non-noise limited system, but, sorry not an interference limited system, but a noise limited system. So performance is much, much better. This is why we started to introduce Massive MIMO. So that is how beamforming would look like. My legacy is only a 2D and in an asimorph. Now with Massive MIMO, you can have elevation beamforming as well, so you suddenly have pencil beams in three dimensions, giving you a huge opportunity of addressing and accessing and connecting users, particularly in urban environments.
5G Unified Air Interface
Let’s talk about the interface, and similar to the spectrum observations, we need something which is unified. And really we are trying to bring together a very different methodologies in here, so we have the different OFDM configurations. We have different frame structures. We need to support advanced technologies, such as massive MIMO millimeter wave. We need to support different duplexing methodologies. Well for that, ideally, we would like to have a single interface, a single set of parameters you could tweak to come up with that 5G setting and accommodate the different technologies. And that’s how we do it, so you look at the flexible radio frame design.
Don’t even know where to start here. But what you see is a time frequency grid. So we have the time going from left to right, and we have the frequency from bottom up into the slot. So let’s start with the frequencies as subcarriers. You will have different subcarriers there just as we had in 4G, but we also introduce black subcarriers because we would like to allow maybe different systems to coexist in between. I’ll leave it for some future type of applications. We have in the time domain different TTIs, transmission time intervals, of different lengths, so again, to accommodate different requirements.
In the frequency domain, we have scalable subcarrier spacing because as you go higher in the frequency bans, you need to make sure that your subcarriers start to be spaced further apart. As they get further spaced apart, your symbol duration goes down, so you then can adapt this by having very short TTIs. So loads of capabilities here where you have a very, very generic time frequency resource grid, which you can use to run very advanced schedulers to make that a very, very powerful interface. And a specific configuration of this all, is called a numerology. So we have different numerologies here, so if you look at it today, you have different waveforms you could use, different subcarrier spacings. So the 15, 30, and 60 kilohertz subcarrier spacing, for anything below six gigahertz, and then higher subcarrier spacing for anything above it. You have different cyclic prefix options, different frame durations, subframe durations, slot options, different duplexing options, physical resource block options, and the maximum number you can have, et cetera, et cetera.
So a very specific choice of all these parameters is called a 5G numerology. This is how we use it, we apply it to the radio stacks. I don’t want to go to too many details, but the three columns you see here is for the three applications, the three use cases we talked about. One is enhanced mobile broadband, video streaming. The IoT stuff and maybe a smart grid, ultra reliable low latency comms. So as you can see on the physical layer, we can adapt to a suitable numerology. For instance, as we go video streaming, we need a lot of bandwidth here. We optimize really in terms of coding. We have a lot of time to do that, so there’s no time constraint per se. And then we can use the traditional mark, regulating control and all the other layers on top. Now let me go to the other end of the spectrum, which is URLLC. In here we really make sure things have to happen very quickly and very reliably. So therefore the radio frame is adapted to be very, very low latency.
We have a TTI which is very short, coding’s optimized, very short payloads really, so we need specific coding for that. At MAC level, we possibly omit the hybrid ARQ. The ARQ is when you get to your signal confirm, your packet confirm, if you get rid of it, then you have of course much less latency. So there’s different ways of handling this today and a lot of flexibility. Of course but the flexibility comes a lot of responsibility and comes a lot of headaches because it’s to configure the system properly is very difficult. So we will have probably artificial intelligence to help us, and be coming to grips with that. But overall, I think it’s a very exciting design approach.
5G Access Network
Let’s talk about the architecture. The radio access network architectures, as we used to call it, we actually got rid of the R because we think the access network architecture is more general when just supporting a specific radio. And you can see on the right-hand side on the bottom side some true 5G base stations. We refer to them as gNB, okay? So these are the 5G base stations. They’re connected between each other with an Xn interface and forming essentially the access network and the gNBs are connected to the 5G core network which goes up on this slide.
Now on the left-hand side you see some base stations which still have eNB in the title, enhanced node b, as we used to call our base stations in 4G. Now they are not any more called purely eNB, they’re called NG-eNB and the reason is is because whilst these are 4G base stations towards the mobile terminal, they are 5G base stations towards the core network. So if a base station is able to connect to a 5G core while they’re still using a 4G interface, we refer to it as an NG-eNB and it is equally connected to my 5G base stations. Let’s zoom in into one of these NG, next generation, radio access network base stations, the gNBs.
And you have different configurations. On the left side you see a single centralized association so you have gNB. It’s a literally an antenna on the roof and it has all the processing in there and it connects not only to its peers but also into the 5G core network. On the right-hand side you see a more exciting development and we call that the cloud run, the distributed unit split. So on the bottom side you see the gNB-DU, the distributed unit and that is my antenna. This is literally my antenna which is on the roof and it may have some functionalities implemented. It could do some of the processing or maybe not. It’s just sampling the spectrum and sending whatever signal that is over the F1 interface down into the central unit, the gNB-CU, the central unit, and that’s my cloud run association and that is where I’m doing all the processing. And I can connect several of these remote radio heads into the central unit and do the processing, and the central unit then connects to the 5G core.
Why do we do that? Well, it turns out that we save a lot of energy. You need to cool your base stations. Typically, you need to provide power, so if you keep everything very light on the antenna side and you do all the heavy processing centralized in a cloud environment, you know, that’s the best way to go. Of course optimum would be to have it, let’s say, in an Amazon web service somewhere, somewhere in a very centralized cloud, but we don’t the time of doing it because we’re talking a latency so we need to bring the cloud to the edge and that’s how we do that.
5G Core and Architecture
That, in my opinion, is actually one of the most exciting developments. Look at this.
This looks now much more like a computing architecture. You remember the computer charts we had? We had a control plane there, which is doing all the control connecting all the control functionalities. And then you had the data plane, where essentially everything happened what the users were doing on the computer. Well, you know what, 5G is now at that point. You see on the top, all the different functionalities and functions and control functions, which are needed to support anything the users want to do in the operator.
We have an application function, unified and data management, policy functions, network exposure functions, authentication functions, so it would authenticate the users, access and mobility management functions, when you take care that you get the right handovers to the right base stations, et cetera, et cetera. And they’re all connected on what you can see here on a data bus, right? And then they control the data plane, which you see on the lower part of the slide where we have the UE, user equipment, which is my mobile terminal talking to the next generation runs, my base stations, which connect then straight into the data network via the user plane function. It’s become a much lighter system and much easier to implement because you can imagine that each of these functions now, could be its own software routine and you can put it on its own virtual machine and you can do with it what you want. And that allows us to go from a very hardware-centric approach to a more software-centric one.
Look at on the left-hand side, this is how we used to be. 4G and before it’s all about boxes. We have base station boxes, we have a serving gateway box, we have the packet gateway box. Now that’s totally replaced now by a set of software components, which run on off the shelf hardware. And you can see that we have the base stations connected to my Cloud RAN, the Cloud RAN is connected to the Core Network, but the Core Network runs essentially a set of servers. You can buy them off the shelf anywhere and you just need to implement the different functionalities. And that allows us to do something now, which is truly amazing, and that has entirely software-ized my 5G system. Look on the lower side, we have the actual access radios, we have 5G, we have 4G, we do have WiFi, anything you want to plug in. Then we have the fully virtualized infrastructure, we have the hardware, the software decoupled and it allows us to do network function virtualization with the VNFs, which is shown on the top left.
We then have SDN. SDN is all about decoupling control plane and user plane. The control plane, if you think of an IP packet as an example, the header is the control plane, and the control part, the actual data which counts as the user part. Why do we split this? Well it’s very interesting to split because we want to support a huge amount of user cases. Think about IOT. IOT requires quite a bit of control, but very little data. Whereas if you want to stream a video, you have a lot of data, comparatively very little control. Keeping that together, control and user data, doesn’t really make sense and place them along SDN infrastructure, and therefore, leverage on the best capabilities of each of these functionalities, wherever and whenever you need them.
This is really the big, big breakthrough in 5G. And this is how network slicing would look like. You see three slices here. One over the other. The top slice is the embb slice, so this is broadband, that watching your Netflix video. Then we have the ultra-reliable latency slice. You know I need to control something in telemetry in the car. And then we have the MMTC’s, the IOT slice, you’re just having a sensors of measuring something. Now you see then also from left to right, the actual equipment connected to the base station, connected to the Edge Cloud, connected to the Core Cloud, where we have the different functionalities.
And I want to explain to you what this slice really means. Let’s look at one specific functionality, one VNF, which is the mobility management one. On the top slice, the mobility management is in the top-right corner in the Core Cloud. It’s very far from the user. It an be very far from the user because the user’s barely moving, we’re sitting in a coffee shop and watching Netflix. At the same time, I may get up at some point and start walking, so we need mobility management, but we don’t need it very close to the user. Let’s keep it where we have a lot of storage and control capabilities, very far in the Core Cloud.
As we move into the ultra-latency slice, we need to support essentially a very quick handover, so my mobility management now sits in the Edge Cloud. Rather than in the end Cloud, it sits in the Edge Cloud. And for the IOT case, sensors don’t move in the application, so therefore we just don’t need this NFE at all. We have just runned it down and we’re not using it. That’s the type of flexibility we get with slicing the architecture. And of course, I’m showing you three slices here with one user per slice. We will have many more slices and many more users per slices and that will give us a lot of challenges, which I think is very, very exciting from an engineering point of view, but I’m sure we will match.
Let’s talk about the migration path. It’s 5G will not come tomorrow. We have a legacy system today, which is our 4G system. And different migration paths have been outlined by 3GPP. Let me talk about one specific one, which I think will happen, and that’s the migration from pure 4G to pure 5G. Today we have an enode 4G base station, which is talking to the in-half packet, called the EPC, which is 4G Core Network. where were have a non-standalone 5G capability, where we have a GNB being anchored essentially to a 4G system, so you can only use the 5G antenna, when a 4G system is around and anchors you from a control point of view. It essentially instructs a mobile phone to listen to a 5G radio signal, but there’s no control going on.
We start introducing the base stations and then we start introducing the 5G core, and that means that my 5G base station can speak natively to the 5G core, but my 4G base station cannot anymore. We therefore need this NGE and B functionality, which I talked about. And once that has been exhausted, then we go into a pure transition. We switch the 4G equipment off and we go purely to GNB 5G base station with a 5G core. When will that happen? We’ll see. Different countries, different raw patterns. Depends on investment. Some countries will do their transition I think rather quickly, maybe in the next two three years. Other countries will take their time. We’re talking maybe 2025, really having a global footprint off a pure standalone 5G system in every single country on the planet.
Author
- Professor in wireless communications at King’s College London.