Open RAN & The Role of The RAN Intelligent Controller

Session Abstract:

A platform where telecom operators can successfully deploy across the entire 5G stack on cloud-native infrastructure at scale, including near real-time network automation and performance optimization, is well underway. Near-real-time RAN Intelligent Controller (RIC) platforms are optimizing performance across operators end-to-end cloud native wireless networks for macro cell, small cell and HetNet coverage. What are the use cases that are delivering improved spectral efficiency of the network, resulting in higher throughput and more capacity, reduced operational expenses through automated configuration and issue resolution, and dynamic balancing of user traffic across the network? Joining this session are Neil McRae, Chief Architect at BT, next is Shravan Vallala, VP of Technology at Robin.io and on the end is Francis Haysom, Principal Analyst at Appledore Research.

Executive Speakers:

Neil McRae - Chief Network Architect, BT
Shravan Vallala - VP of Technology, Robin.io
Francis Haysom - Principal Analyst, Appledore Research

Transcription

OPEN RAN AND THE ROLE OF RAN INTELLIGENCE CONTROLLER

Abe Nejad: A platform where telecom operators can successfully deploy across the entire 5G stack on cloud-native infrastructure at scale, including near real time network automation and performance optimization is well underway. Near real-time RAN intelligent controller or RIC platforms are optimizing performance across operators end to end Cloud-native wireless networks for macro cell, small cell and HetNet coverage.

So what are the use cases that are delivering improved special efficiency of the network, resulting in higher throughput, more capacity, reduced operational expenses through automated configuration and issue resolution and dynamic balancing of user traffic across the network? Joining this session are Neil Mcrae he's chief architect at BT, next to him is Shravan Vallala, vice president of technology at Robin.io and on the end is Francis Haysom, analyst at Appledore Research.

Gentleman Welcome.

Abe Nejad: Thanks for being here. It's been a good couple of years since we've all probably seen each other in person more or less, so it's good to see your faces and I'm definitely glad to have everybody here in Barcelona as that was a long trip for everyone. I think we all have maybe a few more gray hairs, including myself, but it is what it is.

Shravan if you don't mind, I'm going to start with you. What are the primary tenants of really modernizing Open RAN solutions for 4G and 5G networks?

Shravan: One of the main reasons is the cost, traditional ran deployments are based on proprietary, monolithic and legacy architectures. This avoided the operators, not to be able to mix and match solutions that are out there. This increases the cost for the operator and with the microservices architecture now it has enabled automation. It has enabled deployment, stealing healing upgrade, and zero touch provisioning of 5G workloads.

Abe Nejad: Neil, as 5G deploys what can really reduce development time for CSPs?

Neil: Yeah, I mean the key thing is preintegrated solutions. So we see across the open ran spectrum, a lot of different companies building a preintegrated solutions so that we as operators, we don't like complexity. We want buy something, we want to deploy. We don't want to spend years integrating it. We want to get it out there. And we want customers to see the benefit of it. And for us, we really see the benefit of Open RAN. Yeah, of course, if there's any cost savings, thank you, we'll take them. I'm less convinced about that. What I'm convinced about is being able to deploy a better service with better outcomes for customers, and really being able to make a difference in the network we deploy for customers. And I think that preintegrated solution that allows us to hit the road running is really important.

Francis: And I think to Neil's point, I think the challenge is we're changing the environment. We may be changing our services at a very much more rapid tempo than we are used to at the moment. So telco has tended to be, well, we will lab test something. We will work out the whole ecosystem. We have vendors that completely validate every part of the solution. That slightly begins to break down. That doesn't mean you want to throw stuff in and see if it works, but it's a much more agile testing program.

Maybe even some of the aspects of what we were typically tested as a lab experiment. Actually, some of that is coming into the network itself. And we may want to learn some aspects of what cloud is doing, the Netflix chaos monkey type of idea of testing. So it's this idea of bringing testing much more into the ongoing management of a continually changing network.

Abe Nejad: So Francis I'm going to stay with you. So what are some solutions that can improve automation in the realm of 5G and Open RAN? Is there an example you can give us?

Francis: I think that we're at an early stage with Open RAN. We are still very much at the early stage, there are one of potentially two major vendors that are already doing stuff here. So at the moment, we are still feeling our way in this area. I think there's two aspects. One of which is the idea of orchestration. And I don't just simply mean that in terms of aggregating things together, but very much the ongoing management, the scale up scale down of the components, which if we're going to leverage some of the capabilities of RIC, we're going to need that capability.

The other area, which is really key to Open RAN is simply analytics and increasingly AI. Open RAN promises a huge amount of data that is typically to date being hidden underneath EMSS or proprietary interfaces, something you're drinking from a far hose in terms of the data that's available to you. So the analytic solutions, AI, the ability to encapsulate that and turn that into knowledge that can then be programmed into the RIC, will also be a key enabler for success here.

Shravan: And also the integration of continuous integration, continuous deployment and continuous testing would actually improve, reduce the development cycle. So with microservice architecture, you have these C….T pipelines where you can test and deploy and innovate faster.

Abe Nejad: I want to get into the RIC, the RAN Intelligent Controller, but before I do that, something that Francis has just said makes me want to ask you this question, Answer it, however you want. From an operator's perspective, what is the state of play of open ran? We're hearing all kinds of things?

Neil: Yeah, I think the problem is there's no one size fits all. Every network, although we do iPhones and smartphones and broadband dongles, every network's slightly different, different countries, different geographies, different regulations. There are some places where if you're rolling out a brand new network and all you're rolling out is 4g, 5g, open ran, good to go. You can roll it out. If you're like us where we've got 2g, 3g emergency network services, IOT, a whole lot of complexity in the network then is open ran fully ready for that? No, is the answer.

And it's not a bad thing. Let's remember, open ran is still not even a teenager. It's a young technology, it's a new technology, it's going to go through the height curve. It's going to go through everything that technology goes through. But it's got momentum. So we've just done an RFP around small cells and neutral host. And can we deploy that into our network today? Absolutely we can. Could we deploy it into the macro network and to big towns and cities? No, it's not ready for roads.

That doesn't mean to say it's not ready full stop, but it means there's more things that we need, and there's more developments that we need. And in any technology, that's how it starts. It's right for some things, and it grows and gets better and gets augmented. And actually, what the gentleman just talked about doing CICD and more testing means we'll get there quicker than the old model of telco, which is to build a lab and test for years and then roll it out. So I think the future is definitely bright for open ran, but I think where I differ against most of my colleagues is I'm looking at this as a way of generating more demand as opposed to saving money. I'm really unconvinced about the saving money aspect of it. It's about how can we create new services and new capabilities. And you talk about RIC, the RIC really plays into that in my view.

Abe Nejad: We're going to get into that in just a moment. Shravan, did you want to respond to that at all as far as the state of play of Open RAN?

Shravan: No, I agree.

Francis: I was going to say, just drawing upon what Neil was saying, I think the interesting thing about Open RAN is going to be this innovation thing. Let's be frank about it. Open ran is ran. We've done ran for ages. There's a lot of opportunity by, by breaking it up to allow innovation. The most important thing is disaggregation, which is what open ran is about is something that has fired innovation in a whole load of other industries and particularly within the cloud.

So there's huge opportunities, to start thinking about the unknown unknowns. RAN in its current configuration is designed very well. It's a hugely successful system to deliver the type of 2g, 3g 4g and 5g services that we're used to at the moment. But when we start talking about working with enterprises or wanting to add value to the enterprises, the open ran gives you some opportunities to think about, well, what would a ran for a manufacturing plant look like? Which is very, very different than a macro cell network for EE for example. So it's that innovation, that innovation thing is the opportunity here. And maybe operators will deploy open ran, maybe in private networks, managing them for others as their first step there because that's their specialization, that's their differentiation.

Shravan: Yeah. And also the fact that open ran defines the standard interfaces across various ran components. What this enables is for vendors to do faster innovation. So as a vendor, I'm not forced to deploy, create a new ran. I can focus on a very specific use case, on a very specific component, and I can build and innovate and provide a additional service. And then because of this open ran and the interfaces being standardized, now the operator can pick and choose what solution fits a particular use case.

Abe Nejad: So Neil, you started us off on the RIC rather, the Ran Intelligent Controller and how it really plays a role in optimizing performance for CSPs. Put the training wheels on this for us, talk to me like I don't know anything. I know that you guys, you folks do, there are some folks out there that don't. Walk us through that.

Neil: Yeah. In SDN terms, the RIC is a controller. I mean, that's how I look at it. It kind of is the brain of the operation and radio, we kinda need two brains. We need a brain to run the other brain because radio and processing radio is one of the hardest computational tasks there is out there. The only thing that requires more computation is optical services. So you have what we call the real time RIC, which is controlling, like it says almost live, the radio that's out there and making optimizations and changing things and tackling with real-time events, that could be everything from, a bus driving past your radio to re optimizing for a sudden flash crowd.

And then you have non-real time RIC, which is probably where you'll see operators spend initially the most amount of time. And that is deploying the kind of things that we've been deploying in the ran for many years around configuration, around management, around analytics and data. And actually we've just deployed in trial both of these platforms into how, and we're building this up slowly and carefully as we roll out a new network back in the UK.

And already we see optimization benefits coming out of the RIC. Manual or in the old style of building things would've taken us a lot of effort to actually discover. So we are really super excited about this. We've been super excited about controller based technology at BT for years. And now seeing this come to the radio is fantastic. I think the earlier point though is, remember, this is just radio. No one's coming into any of my shops in the UK saying, hey, I want an open ran service, right? They just want it to work. They want it to work well.

And Abe I see our job here in somewhere like Mobile Congress here in Barcelona is I want to increase every single customer, every single user of the networks dependence on the network. That's how I make money. And the Ric is really taking us to a place where we can increase that dependence as mentioned using the Ric to build a network in a factory or in a different sort of environment, where it's not just in the streets, it's about running a business. And I think the Ric really gives us the first opportunity to do that in a way that's manageable, scalable, and that we can make money out of, which for me is really important. So that when 6g comes along, I've got the money to invest in it.

Francis: To just add to that, the world is becoming a more unknown place. The telco is not in control of everything it is doing. It just needs Apple to introduce, in fact, they did this historically, they introduced a version of iOS, which was not good for the network. The network could not react because it behaved in a fixed way. Ric is your ability to be able to intelligently react to new devices, new ways in which those devices are being used. A new social network, a new messaging app or I don't know Pokemon go is an example, completely changes the dynamics of what the ran needs to be delivered. How do you react to that? Telco is not in control of that application, but the Ric gives them the intelligence or the control points of which they could react to that.

Neil: And gives us data, don't forget the data aspect of it. We're able to pull much more data out. Shove it into an AI that can give us a quick view. This is something new that we've not seen before. Do we need to do something? Do we need to alert our operations teams? For me, that's the beauty of it. And I think everything that we're doing in automation, and let's be really clear, the future of the network is automated. Everything that we're doing around automation is powered by that data. And then you get this loop where you feed back into the controller and you're able to make a lot of changes, minute by minute, if need be, but hopefully tackling those bigger issues day by day or week by week.

Abe Nejad: Shravan Neil mentioned near real-time RIC and I know there's non-real time, RIC, again, for the audience, can you either introduce the correlation or the difference?

Shravan: Basically, in the RAN today, there are three control loops. There is the real time control loop, where the time, response times are supposed to be less than 10 milliseconds. Then you have the near real time RIC, where the response times are anywhere between 10 milliseconds to a second. And then you have the non-realtime RIC where the control, the response times are supposed to be greater than a second. So what these allows is to have a hybrid AI in a way, where your near real time ric has certain models that get pushed from the non-real time ric. And it also is collecting data that goes back into the non-real time ric and the non-real time ric has information from multiple ran components that allows us to have more contextual information, so it can make intelligent decisions and train the models and then and push it back. So it's like the standard interfaces that they use, the A1 and the E1 interfaces.

Abe Nejad: You talked about Telco’s being able to intelligently leverage or use applications or technologies that they maybe weren't able to do before. Can you give us a use case or example of leveraging the RIC?

Francis: The one, very specific one, as an example, is very telco related. Vmware, we are working with a company called Cohere, recently demonstrated how you could incorporate something that probably in a previous generation would have being seen as the next generation of the ran. They were able to introduce a capability, which happens to be called reverse Doppler, which is factory anticipating where somebody will be moving based on their patterns to get higher spectral efficiency.

That's something that the Ric enabled to be introduced into a 5g. I forget it may have been a 4g, but a 4g or a 5g network, without the need for an adaption of the standard. So you were seeing a faster innovation path there, a potential to increase spectral efficiency through the introduction of a new, effectively a new tracking of a user using this reverse Doppler effect.

Abe Nejad: Neil, a use case?

Neil: I think the use case carriers are going to be looking at, is very much based on that performance and capacity. That's where our customers complain about performance and capacity costs us money. Where we've been using the ric and trials is very much on getting feedback from the devices that says, actually there's a [18:08 inaudible] spot here, or there's a break in connectivity. And actually that feeds into the ric and the ric can say, okay, I'm going to turn all of these radios. To mask that non-coverage situation so that as you're walking your dog along the street, and you're listening to your favorite podcast, there's no break, you keep going, because find the feedback from the user that there's an interruption there.

And we can act on that in real time. And that just increases our dependence upon the network. And we can take that example into some really high demand scenarios, like an automated factory. It's probably some way off, but I can see us building more and more of a network to control that automation and provide the kind of right safety levels and the right control levels so that someone who's building or making something can do it in an automated way and get real feedback through the network, because we're able to control and manage and provide a network at a level of quality that we've never really been able to do before.

Francis: And then another area maybe in terms of just basic energy consumption, dealing with title effects, for example, like lots of people are using a motorway in the morning, the rush hour. They're not using a motorway. You may want to completely change the way in which handover is managed. The direction of tilt depending on the time of day, based on the traffic and that traffic analysis. And as a result of that, be able to turn on or turn off capability and not consume energy for example. There's a huge number of variations of that kind of theme of the ability to just have a network which actually just breathes.

Neil: Yeah. I mean, I think you're right. Energy is a great example. And I think as the price of energy increases and the demand for more renewables, I think it's going to be one of the biggest use cases.

Abe Nejad: Shravan, from your perspective when we talk about the Ric, aside from the use cases, of course, let's say a year from now, we'll be talking about more applications and more use cases that the RIC can be leveraged with. But aside from that, this time next year, what do you think we'll be talking about when we talk about the Ric?

Shravan: So as Neil mentioned, right, Ric is basically an SDN controller and what they have done is they have enabled the ability to deploy these various applications, which they call [20:35 inaudible] and RX, which are going to pave way to newer things that we have not thought about today. For example, like some of the things that I can think of is what Francis mentioned, where you can credit the usage patterns. And based on that, you could reduce the clock frequency of your CPU, that's saving power, or you feed these data into an orchestrator, which can move the components around and make sure that the workloads are consolidated into a smaller number of nodes so that you can shut down the remaining nodes, all this and more is going to come.

And there is also service assurance, where today the service assurance is based on user groups. There is a possibility where the external information like GPS location of a first responder can be fed into the non-real time ric, and then at a user level, you can provide priority across your network.

Neil: And absolutely, that's actually a real use case we've been trialing in Birmingham here in the UK, which is the connected ambulance. So if you imagine an ambulance has been at a road traffic accident, someone's injured during the ambulance. The ambulance is armed with cameras and sensors. It takes this picture and through the streets that the ambulance is going take, the ric is optimizing the connectivity. So it can upload a really high definition picture of a broken leg or an MRI so that when the patient arrives at the hospital, the surgeon's already got a good idea of what the patient's diagnosis might be. That's a real use case that we are trialing right now.

Francis: But to your point, it requires you to optimize for an ambulance high availability, high quality, good handover, no dropped call, as opposed to say on the same network, an IOT device that is fixed or placed needs no handover. And in fact, doesn't even need to be resilient because if I fail to connect once, then I'll get the recording another time, completely different way in which you'll be using the Ran efficiency, the spectrum of the Ran. And at the moment you have embedded in integrated solutions, you kind of have a one size fits all story for every use case.

RIC is your ability to break out of those use cases and say the IOT device needs a very different set of applications and handover management and the rest than Neil's ambulance rushing to the hospital.

Abe Nejad: Yeah. Interesting. We've run out of time unfortunately, I know the Ric will be a topic that people are discussing here in Barcelona, and it will be going on for quite sometime. So by the way Francis I'll ask you this question, has the RIC been around longer than we think? And it now just has the acronym around it and people are talking about it. What was the impetus for the RIC?

Francis: I think to Neil's earlier statement at the beginning, the Ric is a controller just like SDN. It's the idea that I take the control out of what is an embedded device, and I allow it to become software program. That's been going on for a long time. Really the key thing about the ran has become a heavy compute, it's a very difficult problem to do. And we've tended to build that in hardware today, it's been encapsulated because that's the only thing that can deliver it.

Open ran is really saying actually some of that could be on commodity hardware, major parts of the way I control the ran could be using accelerators for example, but I can make that programmable, it's that jump of making it software that enables the ric. The idea is controllers exist already in a typical network vendors solution. It's just that you can't get at it. And what it does is totally dependent on what the net will provide for you.

Abe Nejad: Okay. Understood. Well, again thank you all for your time. I know it's towards the end of the day, on the first day here of the event in Barcelona. So we appreciate everybody's time. I know you have probably somewhere to go. Francis, we haven't done this before, but thanks for being here. I know we have another one of your colleagues, from Appledore that's going to participate. So thanks again for that. Shravan, you did back to back sessions with us so you're probably exhausted. I know the feeling. I've done nine, but I don't have to answer the questions, I only have to ask them. So thank you so much for your time and your effort.

Neil as always first of all, it's good to have an operator’s perspective. Second, it's good to have you on.

Neil: Well, it's great to be back in Barcelona. If you're at home watching this, you should have come here because this show's been great. It's been fantastic.

Abd Nejad: Yeah and it's just the first day. So again, thanks for everybody's time.

Thank you. Thank you.

###

For any inquiries, please email anejad@thenetworkmediagroup.com

Abe NejadMarch 14, 20225G RAN