Stefan Parkvall - 6G – the intelligent network platform of 2030

Stefan Parkvall - 6G – the intelligent network platform of 2030

Show Video

thank you emil so welcome a few of you that are here fiscally and the vast majority of you that are remotely participating so i will try to talk a little bit about 60 and try to give a little bit of an answer to this question what is 60 i to shed some light on that and i will follow this structure to start with a little bit of driving forces go through use case capabilities and then come into technology and towards the end also talk a little bit about the timeline when we foresee this will happen more more from the standardization perspective but we just have to see we're basically talking about something that will be deployed in 2030 and onwards so it's not really 10 years from now but but almost that's a kind of time perspective so let's look at some of the driving forces what could drive the need for for a future system and we have identified four of these forces first one being application demands with that we basically mean existing applications as well as new ones that we don't know about yet what they needs they have in terms of day trades lately capacity and so on so they will of course push for more then we have trustworthiness and that pc has something very very important if you're going to connect everything the society will depend on this system it needs to be trustworthy and historically i think seller systems has been and still are very trustworthy there have a lot of cyphering encryption and so on but if you broaden the the space that you try to cover with 60 go try to ground connectivity or include compute and other things we also need to broaden the concept of trustworthiness that you should be able to trust that your data stays where it is supposed or that your your compute and connectivity is always present never fails and so on there's a very broad broad thing not just the encryption third fourth we have here is sustainable world and i think that goes without saying the world is not sustainable the way we run it today and we need to change that and how can 60 and also 5g help with that the last panel it's called simplified live here on the slide or maybe not the best title but its intention is to capture the distributed ai that we received we will today see a lot of artificial intelligence and machine learning and things like that running and helping us to to do different tasks in society that will become more and more common in the future and these ai machines need to talk to each other and they would have demands of course in terms of data rate but that will end up under blue bubble but also other demands what type of data forms do they exchange information with how do you build the distributed ai machine and so on that will be the fourth strike so these are the four things that i've seen that come from the outside perspective that could drive the need for something something new and then use cases and when i took use cases i usually say that whatever i say is probably dead wrong because it's extremely difficult to predict use cases in detail i may remember the 3d killer use case as well video telephony no one did that but it happened later on with iphone and androids or so on so it's very difficult to pinpoint a very single specific use case but said just brought in this slide trying to illustrate a little bit what we have in mind and it is a lot about bridging this digital world and the physical world it's grabbed the fiscal world where we live even though it's been a lot of digital in the last two years the fiscal world where things happen we have sensors we collect information bbm affect things and every digital world we did it's a twin very detailed model of that physical world where we can do experiments we can try out new things try to figure out what to do and then feedback that information to the digital world so it's a lot about moving back and forth between these different verbs so we have connected into your machines for example machines that connect and do things together it could be robots who could be a robot and human to work together in a task or something like that internet of senses in something that's just an extension what we have done in the past i mean the first telephony 1g that was really remote audio that you can hear things later on we added video but why should we stop there why not continue with touch and smell and a lot of other things and then this digitalized and programmer where that's a lot about going back and forth between these two domains and of course we should do this in a sustainable way actually we will save my reputation of predicting the wrong thing by not going into details and use cases we can try to do that we can have a chat on that afterwards if we're interested but i'll just leave it here for also from the interest of saving some time and the next step when we know a little bit about use cases or at least what type of use case we have what capabilities do we need so let's start with what i call the classical capabilities in in this blue circle uh for example has achievable data rates of course we want higher date rates we would like to transfer information faster in the future we should be a little bit careful though we should not talk so much about peak race the there's a tendency a lot of times to end up with a peak rate discussion but the peak rate is quite irrelevant what really matters is that she will redirect what data do you actually get in reality but we want higher achievable date rates same thing with latency of course we would like to push down latency inside we talked a lot about one millisecond and turned latency there are cases when when you wanted to push that further down but probably more important is to have reduced jitter many applications run just fine with one two three milliseconds but they depend on very legitimate so it's probably more important to enhance in that domain than in the absolute lowest number positioning that we've had for a long time and that's something important being able to position things where they are traffic capacity another classical metric have been with their cnc 1g we always need more capacity because people use this system more and more and you can find in there to mobility reports how do the traffic need is increasing we need to meet that with more capacity coverage of course if you don't have coverage you don't care about the restaurant metrics because then you you don't have a connectivity so coverage is also very important and really truly global coverage everywhere would be nice to get and network energy efficiency our systems need to be sustainable and cost efficient and for that they need among other things be very efficient with energies we need to push that down but all this is in some sense classical capabilities that we talked about also in the previous years they continue to be important but we should not view 60 just as 5d times some factor of x we should not just scale up five we should do something more so we actually have an outer ring of um i call it new capabilities you can of course discuss what's new what's classical under the border between the two circles it's a little bit blurry but trying to address new capabilities for new use cases security privacy that goes back to trustworthiness in a wider sense not just for connectivity but for everything service availability trying to do the service of course needs to be available there i need to trust that my my service up and running even if someone accidentally cuts off the cable to the base station or something happens then you still would like to run your service at at least to some extent for example if you have connected your your factory with 60 and suddenly there are some excavated droughts in the street like casa cable you don't want your factory to stop it still needs to operate and you continue to function service versatility you would like to do maybe deploy a lot of different types of services in a flexible way and scale them up and down dependable compute with that we tried to address the fact that 60 is more than just communication it's also compute and some of that compute tasks they may execute in the in the device some of it may execute in the central cloud or somewhat in the edge cloud and depending on where you are and what needs to have this compute task may move between these different places in the network and to do so you of course need interfaces and apis and mechanisms to be able to move that compute test back and forth that's one example we need to do that deployment flexibility i think it speaks for itself we want to be flexible in our deployments run in different types of spectrum applications be able to set up networks temporarily if you have a rock concept or disaster zone or whatever sensing that's an example of a new capability that in some sense is an extension of our positioning but really what we have in mind here is that if you anyway deploy a large number of base stations and other nodes for communication purposes can we reduce that infrastructure also to sense things sense the surroundings a little bit like a radar for example trying to build up a map what's around us if we can do so we which we can yeah and i require some of that then we can feed that information into the digital twin and and know more about what's going on and build up a more more detailed twin and the last example of a new capability is extreme devices in particular for very low-end iot applications there are a lot of applications where you really want to super low cost of your device very low power consumption way below what we can do today we're doing with today's iot standards so that's one example of an extreme device and then last but not least we should not forget cost efficiency and sustainable it needs to be sustainable both from a cost perspective but also for in general to be the friendly to the planet and as an engineer of course you can come up with a lot of cool technology and fans and mimo schemes and whatnot but if the cost of those new schemes new things is too high compared to the benefits they bring that won't happen even if it's funded with the technology and it truly is you should keep this custom and the sustainability in mind as well so that it really can happen in reality so that's a little bit about capabilities and the bottom line here is that 60 is not just a scaled up 5d it's something more it's broader so it needs to have broader capabilities and also it carries more than just what would try to cover and then technology we have a group on the slide i grouped it into four areas so i thought that took away the animations okay first one being limitless connectivity so that all types of technology that can help us with we're getting a truly limitless connectivity the division here is that no matter where you are and what you want to do connectivity should not prevent you from doing what you want to do if you have a certain application certain task you want to do something and you need a megabit for that then you should get a megabit if you need a gig of it then it should get gigabit and so on the connectivity should only be there the connectivity should not stop you from doing what you want to do that's the division and then we have a bunch of technology groups here and subgroups and technology that will help us net we'll get back to some of these in common slides we also have trustworthy systems and that goes back to this trustworthiness we talked about the beginning and it's a little as i said a broad range it's not just for the communication part it's also for a compute part how do you run this kind of cluster enclaves in the cloud so that your information doesn't leak to someone else with all of these aspects how do you trust that that's the one you're talking to is the right one and you need some trust certification mechanism and so on so everything that's do you have to do with trust we have an error that we call cognitive networks uh also that was divided into a number of smaller sub areas but the cognitive networks we really mean the idea of using ai and machine learning inside the network to optimize how you run the network both the communication part and but also the compute part of it so ideally what you should be able to do is to have some kind of intent based configuration of your system you should be able to tell your system that i would like to set up an iut service that should have this data rate and it should not affect my existing mbp service for example and then the cognitive system should be able to figure out how to configure all the parameters how to set up the system to run this new service should learn from what's going on the data that's flowing through the system and there's compute and all the processing that takes place and readjust parameters and optimize as we go and also be able to explain to me as an operator that i took this and this and this action because of so that you can really configure the system in a high level manner just tell you intent through system the system comes back and say yes i did this and this and this because of that's that's the id and use a lot of ai inside the system and great data driven operation and things like that and the last area is natural compute fabric um i think maybe it's been um you know this from what i've said so far and from what they have elsewhere but we really see that communication and then computing is going together and kind of mixing so we will be running parts of our communication networks on on cloud and if that same cloud is also running the some other applications and then we have this kind of unified communication and computer infrastructure and that that we collected here on the network compute fabric so there's a lot of tubes how we can unify these two worlds and then what mechanisms you need to make that happen that's kind of four major areas of the technology that we have in mind for for 16. and then of course as i said within each of these boxes there's a lot of things to do and think about i will just have a few examples on this first in a few slides because what really what we try to get to is some form of intelligent network platform rather all the so say classical networks cellular networks at the bottom they will evolve of course background battery 360. and we have some form of platform that can integrate these different networks maybe add some some value added functions and provide a unified api unified interface up to the application developers so they don't need to bother with each individual network because if you develop an application you you just would like to focus on that application develop it once and it should run everywhere no matter the details on the communication network so that is a little bit of what i try to illustrate in this middle platform here but this is something we don't have to today i would say but you could of course plug in 5g and 4g under here as well so it's a kind of gradual path into that domain and then on top of that you can deploy all your your application whatever that is maybe these examples are a little bit 5g ish but nevertheless if you come up with something new here just put it up there so that's a little bit difficult about the platform that you're able to come up with one platform that supports your application everywhere no matter where you are okay let's jump a little bit into technology and here's just a few examples of technology components how do we get back to most of them in the coming slides so i will spend more time on this slide you start on the i'll call it a network adaptability or but it has really to do with this cloud native things so what we see as i said we see that a lot of communication they should be able to exploit and benefit from the the cloud the general community compute cloud stuff we have today all those tools we have uh fully service based that has to do for those who are a little bit more into the details on 5g you may know that the 5d core is more of a service based architecture so basically instead of having nodes to talk to each other you have function functionality that that pretty much any functionality in the core can talk to any other functionality so it did did that in the 5d core but they haven't done it on the ram side but we see parts of that will come down to to to run as well and then have a keyboard some function separation and the simplification and that that really has to do with the fact that this is a very complex machinery we should be careful here to make sure that it's doesn't get overly complex we should really think through what kind of interfaces do we need to standardize in order to get the interoperability obviously between the handset and base station but maybe we shouldn't try to decompose and chop up all the different pieces inside the networking in very small small units and standardize them because then it will take a very long time be very complex so try to keep it simple as much programmable devices and network that also relates back a little bit to this use of ai inside the networks but a few examples of what one could do here and what we want to do is to be able to deploy new functionality in the network without having to go through the standardization process which obviously takes some time so a little bit more of this devops like style that you run on the height industry to try to bring that more into the compute side as well and one way of doing that is to have the network and the devices more programmable so you can think of about training an ai model on the network side with all the compute facilities you have you know access to all information you had once you have trained it you can download it into the device and execute it in the device so for example if you come up with some new handover mechanism that you think is much better in your deployment you have trained that to develop that you can download that in the device and the device can execute parts of of that handover procedure in a new way instead of having to go through standardization and then come up with something new that which takes time and then flexible and dynamic networks for example we integrate new gear types of access nodes you see a small satellites in the illustration there for example so how can we include satellites into this overall network as a compliment because if you're serious about having a truly global coverage there are places on earth where it's difficult to put up a base station even if it would like to do so oceans for example so satellites could be a complement in some areas and how do we integrate that and how to interact with that in internetwork these are a couple of things what you have in mind on the modern architecture side so the same then we jump into radio spectrum spectrum is really essential we can't do much radio communication without spectrum and what we say about 60 is that it should cover wide range of spectrum from sub degrees up to what you can call sub terahertz it's even wider than than 5d that said not all parts of these frequency axis are equally important these lower frequencies here one two three four gigahertz they will remain extremely important also in the 60 era as they are good for providing coverage it's very difficult to provide wide area coverage if you go up to several hundred gigahertz because the radio propagation is very tough up there so for coverage region reasons you still need to use these lower frequent bands that's fine the um has been one need to take into counter is that there is already cellular networks running in this frequency range 40 and 5d for example and they won't disappear overnight so if an operator would like to introduce 16 it needs to be introduced in what i call the spectrum compatible way you need to to have a 60 that in one way or another is able to operate on top of a 5dn40 carrier we actually did that when we introduced 5d we had the mechanism that the 5d carrier can be put on top of 40k and they kind of can coexist i think we can do it even better than what we did in in the 5d case so let's try to do that in 16 by learning from what we did in 5g and that would allow an operator to gradually introduce 60 in this lower spectrum where they already have 4g and 5b running a very important piece of spectrum and very important that we can introduce it in a smooth way another range that is interesting of course between these two red boxes is already irrelevant but the 7 to 24 gigahertz range we see today norris is going up to six at least fundamental is nothing in the technology that would stop us to go higher in this low high fusion spectrum but the 7 to 24 gigahertz range is currently not used by 5g and it's a very nice frequency range it's low enough to give decent coverage so of course it would be nice to run 60 in this range as well however there is a couple of other users of this spectrum at the moment it's radio links it's some satellite communications and so on so to get access to this spectrum we probably need to think how can we coexist and how can we share it with those existing servers in a smooth way and then we have 30 gigahertz and a bit upwards the millimeter wave range that's a pioneered by 5d that will remain important also in 60. so that that will be there but then basically a comment on the subterror's range because quite often if you go to 60 conference you hear a lot of people talking about these terahertz frequencies or the sub turrets frequencies and yes it's a new frequency it's of course very challenging and interesting from an engineering perspective it's quite tough to be a lot of components up there it's they're not as mature as they are further down however the propagation conditions up here is also very challenging you will not be able to go around corners you will have difficulties going through a window and so on so these very high frequencies they can offer a lot of bandwidth you can provide high date rates but only in very specific scenarios so i think there will be more for kind of niche applications important range is still the slower range to seven to 24 in the millimeter range the terahertz range is more for niche applications but 60 as such we will support all these frequencies not just the red ones that i highlighted here then one of the other technologies that i had on one of the earlier slides that was what we call the sierra energy devices and that's just an example of an um device that you don't have to replace the battery so it's kind of a battery less from an end user's perspective you just deploy it it does what it's supposed to do you don't have to go there to change the battery or to charge the battery it can harvest energy from some other sources could be temperature gradients could be vibrations uh solar power in some cases even the rf waves themselves so no need to change battery and that's of course quite attractive if you think about some sensors or something like that that you can just put out and forget about them they just sit there and do whatever they need to do connect to temporary information or whatever you wanted to sense and feed that bank so this is the type of devices that is much more extreme than the current iot devices in cellular networks to narrowband id and ltm because this is very very as you see in the figure very low down in the power consumption and complexity scale this makes it quite challenging to design because with this tiny tiny amount of energy you have available what kind of fiscal air should they have how should you do mobility how should you do security and sci-fi encryption and things like that you need to rethink a lot about the the radio and the protocols here if you're going to support these cranial devices so this is quite an interesting area actually and you can also envision that in the long run maybe you can combine this with printable electronics you can just print your your 60 low end device on a piece of paper put somewhere maybe on a milk cartoon or milk box or whatever just to track the temperature and then through the industry and things like that has the tracking in the in the factory hall and things like that same principle you can replace barcodes which i'll use today with this but the barcodes you need to see to be able to scan with this you can basically have a pile of boxes on the floor and you can read each individual box and see what it is some maybe even the temperature inside that box or whatever you want to read so that's the application here but that will require some new rage access work as well another new area would be joint communication and sensing if anyway deploy uh communication infrastructure for communication purposes why not try to use it also for sensing purposes so that we can try to sense the surroundings collect information what's going on around us and so on and that the information i collect could be different types it could for example be if you have anything that affects the radio rates basically i mean we have already shown from an ipsy perspective that you can actually see whether that you have low rain on a very local scale or not by just looking at the signal strength on microwave links that's of course very useful input to ready forecasting services but you can do more you can use them as some kind of a radar type of setup so maybe you can track cars that are moving around you can see whether the room is empty or full of people there you can actually in some scenario see where the heart is beating or not on a person with the challenge here is of course a little bit to see what type of things can be can be sensed and what are the things we can sense make sense to sense so to say and that that can be useful in reality but i think one should keep in mind that because sometimes you see that people say yes yes let's go up in the terahertz range we get get an excellent radar probably true but you don't get excellent communication and then you kind of lose this joint communication of something so i think at least the way i look up on this is that we start from a communication perspective and then we try to expand into the sensing and try to reduce our communication equipment also for sensing and that fact really would mean that we need to also handle the frequency ranges that are actually used for communication also the type of infrastructure we put out there for communication when it comes to requirements on power amplifiers and things like that so it's really joint communication something because then we can get the sensing halfway for free compared to building a separate sensing network where you have to pay for that network in itself and at the beginning i promise a little bit on the timeline when will this happen under this is a timeline from a standardization perspective we're in the beginning of 2022 today and we predict uh not just we it seems to be quite a wide industry consensus on this timeline that in 2425 there will be discussions in cbp on requirements for 16. what should the 60 system be able to do what type of data rates what sensing accuracy etc etc and then once that is done in 25 26 and up to 28 or so there will be this actual standardization work to come up with the first standard for 60 that hopefully is ready in 28. you probably see some early commercial problems in the 28th time frame a little bit on wider scale in 2030 or so so that's the the timeline we see but that means that we have a roughly two years from now until we start to discuss requirements in 3dpp and maybe three years before we start to discuss technology there is still a little bit in in the research phase for 16 but it's not so that it's a next huge amount of time it's a couple of years before we start to see this happening standards and then we have the whole standard procedure to run through of course it takes another couple of years but that's a little bit how we see it on the standard perspective that actually brings me to the last slide a little bit ahead of schedule perhaps but since we didn't have questions during the the presentation might be good at some time at the end we talked about when that is this is the 2030 time frame you're talking about we said that 60 is more than just communication not just the scaled up 5d system it's more indeed includes the compute it's really a platform that that provides is this trusted ever-present communication that also was seen in the capabilities we talked about that we had these two circles and i said the outer circle are our new capabilities that i haven't really talked about in in previous g's but will become very important in 60 era and then i mentioned a few examples on technology components on on ray designs it was not the complete list there's a lot more than i would add in slides here but just to give you a little bit of flavor for for some new some of the new things i took a few couple of examples so with that i stop and i guess we have plenty of time for questions at least 20 minutes or so

2022-05-08 18:07

Show Video

Other news