Open and Smart RAN: Deployment Scenarios
CommTech Brief subscribers receive session directly, on August 12th. Sign up for Daily CommTech Brief here.
ORAN has undergone a series of progress from building open architectures and interfaces, to introducing rich capabilities for O-RAN in commercial networks. The next steps are deployment scenarios and the utilization of machine learning and artificial intelligence, via open and standardized interfaces in a multi-vendor network; As well as testing and the integration of multi-vendor ORAN solutions, through end-to-end test specifications. Watch this executive engagement with leaders in the ORAN sector, on the current state of play for Open and Smart RAN, and what to expect in 2021/2022.
Executive Speakers:
Caroline Chan, VP and GM, Network Business Incubator Division, Intel
Patrick Lopez, Global VP Product Management -5G, NEC
Neil McRae, Group Chief Architect, BT
Program [Full] Transcription:
Abe Nejad: And ORAN has undergone a series of progress from building open architectures and interfaces to introducing rich capabilities for ORAN that in commercial networks. The next steps are deployment scenarios and the utilization of machine learning and AI via open and standardize interfaces and a multi-vendor network, as well as testing and the integration of multi-vendor ORAN solutions through end-to-end test specifications. Today, we have leaders in the ORAN sector on. The current state of play for Open and Smart RAN and what to expect in 2021 and also 2022. Joining us are Caroline Chan. She's Vice President and General Manager of the 5G Infrastructure Division that at Intel. We also have Neil McRae, he's Group Chief Architect that at BT and last we have Patrick Lopez. He's Global Vice President of Product Management for 5G at NEC Corporation. And welcome to the program.
Patrick Lopez: Thank you.
Caroline Chan: Thank you for having us.
Neil McRae: Thanks for being here. Thanks for coming back. Anyway, we had most of you not too long ago talking about ORAN and sort of the state of play at that time. So it's good to have you back again. Caroline, if you don't mind, I'm going to start with you. So what is the state of play for moving towards an open and really smart RAN to build a more cost-effective network through these open interfaces and open hardware?
Caroline: I think the journey definitely has started. We are seeing the [inaudible01:33] being made available. We're seeing testing. We're seeing portfolio, product have been announced. Like we are seeing announcement like ORAN Intel in Germany by Deutsche Telekom, [unclear01:47] and ORAN Alliance have all announced testing plants. Definitely the fact that we are here talking about this is a great milestone. We've always been saying things like, oh, this is going to happen, but finally it's here and it's happening. The fact that you have this aggregator power that separate from the software allows innovation to happen. Things like AI and machine learning is happening at this pace that is not bound to the hour, which is really what we've seen all along in the enterprise, in the call center. So it's finally happening in the networking and US space as well.
Abe: And Neil and Patrick, anything to add?
Patrick: Sure. Absolutely. Completely what Caroline was alluding to. At NEC, we've seen Open RAN emerged from a concept to a reality. You might know that NEC has deployed some of the world's first Open RAN deployment at scale in Japan, first with NTT DoCoMo and then with Rakuten. And now we've those deployments expand outside of Japan with announcements from Vodafone, from Telefonica, from Deutsche Telekom. BT is also very active in that space. And we're seeing that from being an idea, Open RAN now has come to a point where it has been deployed in urban environments with massive MIMOs or high performance products. So it's really fulfilling basically the promises of providing an ecosystem for 5G that is open and desegregated as well as automated. So it's really an exciting time.
Abe: And Neil the state of play for Open and Smart RAN?
Neil: Yeah. I mean, honestly, I think if you'd stepped back two years and said, "Where would we be two years from now?" I think we're way further forward than anyone could have imagined. The pace of standardization, the pace of collaboration is just being out of as well. Probably the fastest thing I've seen moved since the early days of open stack and back in the early cloud days and I think, you know, whilst the pace has been good, there's still a long journey ahead of us. You know, I think there's still much more to do in macro and super macro. And the reason we go into Open RAN and O-Ran as it was known before, you know, Ultra MIMO, which is really as Patrick said, using those extra capabilities to get a lot more out of a spectrum that many operators like ourselves have to invest in. So great progress, fantastic progress, still a lot to do.
Abe: So Patrick, back to you. So how are you introducing rich capability of ORAN architecture in these commercial networks, really for these selected use cases?
Patrick: Well as mentioned Open RAN is starting to be implemented in a number of scenarios, a number of use cases. I think it starts in many countries from an experimental standpoint, in a rural environment as an alternative basically to providing deployment at large scale, lower costs. And that's important because basically at a high level RAN, traditional RAN systems have been designed primarily for urban environments. So when it comes to deploying them in rural environments they might not be well adapted or they might be just very expensive for the need that you have, which is basically a high coverage with low density area. Open RAN allows to have basically fit for purpose deployment because it does have a larger ecosystem. You have more vendors and some of those vendors have various specialized equipment, whether in the hardware or in the software. So you can eventually choose what kind of radio solution and what kind of connectivity products you can deploy work.
So that's started with rural and then we've seen moves also in the private 5G environment. In Japan was deployed I think half a dozen private 5G environments for industrial IOT, for transports infrastructure as well as for the tourism industry. And we see also that those verticals have very specific needs in term of connectivity. Needs that either were not being addressed with the traditional RAN or were just not economically viable with traditional RAN. And now we see basically Open RAN moving into the mainstream RAN deployment where traditional vendors have been the one that had been dominating. And we see that basically the same benefits that we see in other areas are becoming more and more important in the traditional RAN ecosystem as well. Benefits of openness being able to pick and choose options in terms of vendors, in terms of technology, in terms of use case that fits the purpose. Obviously, being able to have lower cost of ownership because a lot of the Open RAN solutions end up being cloud-native, which allows for a much faster implementation deployment and lifecycle management as well as basically being able to have a very cloud-oriented security, purpose, and automation. So all those benefits we're seeing little by little being deployed in RAN environments using Open RAN technology.
Abe: And Caroline, are ORAN in commercial use cases?
Caroline: Yeah. So to add to what Patrick said. So besides rural, actually here in the United States, there will be a launch in Fall with Dish and you will start in an urban area, a high dense area. That's one. So it's very versatile. And in addition to some of the private 5G we are seeing, for example, we have been doing Smart Education, given all the remote learning that our children are facing. We are rolling out Open RAN based solution to cover specific areas around elementary school, announced one in Sacramento, California. We are rolling that out because of the fact that it is segregated. It does allow us to innovate at a different pace. So the blueprint itself is going to roll out in many different school districts. We're also looking at some of the things we can do in a farm. We [inaudible08:55] in the [inaudible08:57] County, Washington State are taking the Smart RAN at the farm environment, providing sensors and collecting data and really giving us a most vital part of our society, food, security of food. So the fact that it is that smart, it is open, it really allows us to scale into many different use cases.
Abe: And last Neil, ORAN in commercial networks or use cases?
Neil: Yeah. I mean, Caroline's absolutely right. I think a lot has been done in niche areas, but we're starting to see it go into the mainstream. So we have a number of use cases. One is for a neutral host in a science park. We actually have been working on COVID vaccinations and research here in the north of the UK. We're also starting to see Open RAN into deployments like stadiums. And the thing we're really excited about is the RAN intelligent controller. You know being able to change the capabilities of the network in real time. So, you know, right now where we sponsor Wembley Stadium and at the moment is empty, so we turned the power down and we use less energy, which is really important from a green and sustainability point of view. When the crowd's in and we turn the power up and everyone gets great capabilities and we're able to optimize the signal.
You know, we're able to optimize a signal much better for transportation where I think it's a key frustration for customers. And I think for me, that's what's really important is how does the end user see the benefits of Open RAN? And I think we've been too focused on some of the internal kind of engine, let's think about the customers. Let's think about how it benefits them. And we certainly see, and what we've done in private networks, what we've done in neutrals and what we're doing in the interim, microcells is a big win for customers. Because let's be honest with you. We have the biggest Open RAN lab in the UK, probably in Europe. I've got 500 shops where our customers come in and none of them are asking for ORAN. They just want excellent mobile capabilities. And I think ORAN has taken us from our me too, into a place where we can actually add much more value as an individual operator.
Abe: Neil, I'm going to stay with you on that. We talked at the top of the program about machine learning and AI, that we would be discussing that during this session. So what are some ORAN use cases and deployment scenarios that really utilize machine learning and artificial intelligence modules?
Neil: Yeah. I mean, let me build on the stadium example. You know, you get a lot of different user based directions in stadiums. You know, from as people come into the stadium, how do you optimize the signal outside so that everyone's got their online tickets. They can scan in really easily. They then go through the middle of the stadium, they're buying a beer, they're buying a hot dog, ensuring that they're able to kind of communicate and take pictures and really again, reaming that signal. And then when they get into the stadium, you know, they're watching the soccer here in the UK, we're able to look up what the stats are on their favorite player. They're able to take a picture and send it to their mom. And all of that in the past was really hard to do. With this capability, it's getting easier and easier. It's still difficult, but it's getting a lot easier for us to really match the experience customers want versus our ability to provide it.
And frankly Abe, humans can't do this. It's beyond our ability to go into a place like Wembley Stadium. It's a hundred thousand seater stadium and adjust radios manually. It's just those days are gone. So what Open RAN and the REC really allows us to do is position things in a way that gives us the best experience for customers tailored to that moment. And also gives us an anomaly detection when things aren't quite right. It will flag and say, " Hey, we don't think anything's quite right here. You should probably check this out or something strange is going on." And perhaps we've got an interferer or perhaps we've got a power problem and that real machine learning and AI allows us to pinpoint any issues very quickly and deal with them instantaneously. And in something like soccer, you know, when we just had the Euros here in Europe, you know, you can have goals happen in seconds, and if you've missed it because you've got a radio problem, that's very frustrating for the public. And of course, broadcasters are using 5G more and more to actually shoot live programs. And we run BT Sport in the UK and we're experimenting with 5G and ORAN technology to bring the best action to our viewers.
Abe: So Caroline or Patrick, let's start with you. Deployment scenarios, use cases that utilize machine learning and AI?
Patrick: Well, I think as Neil was mentioning the networks are becoming more open and disaggregated, which means that there are more and more elements. At the same time, there are becoming more and more cloud-native which means that these are all no longer, you know, boxes with software. But we're talking about a cloud now with a lot of mutualized elements which means that the network changes. It changes all the time, actually. And in order to manage that change, you need to have a fairly complex management system that is able to detect in real time what are the conditions of the network, but also as Neil was mentioning, what are the needs from the customer, basically. In the past, the only to manage something, a situation like the Wembley Stadium was to basically bring as much capacity as you could for as many people as you could, and hope that you will be able to serve that capacity when the time is needed. And that in some cases was a lot of waste because you would build a lot of capacity in a very small space for a very small period of time. And then either it would stay there unused, or you would dismantle it, but it would be very expensive basically to manage.
Now with cloud native networks, you have actually the capacity to allocate capacity as needed and also to target. So one elements of what was mentioned here is enabled by technologies such as [unclear15:40] and [unclear15:41] where in the past, I mean, an antenna was one antenna. Now an antenna is composed of a lot of antenna elements. So our latest antennas are 156 antenna elements. And basically what happens is that the antenna detects how many connections are being made with that antenna, how many people basically are using the service and what kind of service they're using. Are they streaming video, upstreaming or down streaming? Are they uploading, are they reading emails? Based on all those information, the antenna is able to programmatically focus the capacity of that cell to the specific users that have the most needs.
So what this means is that it provides a better user experience for the people who need it at the moment they need it without necessarily having to incur the additional cost of over provisioning the network for this whole stadium. So stadium is one example. We spoke a little bit about industrial IOT, that's another example where in manufacturing plants basically you have robots’ devices that have very different connectivity needs than a person might have. They might not be needing to uploading videos or browsing social media, but those devices, those robots, in many cases, they need to be able to collaborate with each other in real time. So you don't want the robot to hit another robot and you want the robots to be able to perform a very sophisticated task. In order to do that, you need to have a network that is very well calibrated in terms of latency and in term of managing basically all the different movements of all those different elements. And the only way to do that really is to do that with a software that is cloud native that allows the capacity to synchronize and manage the movement of all those robots and Open RAN enables that natively, if you will. So that's kind of an added benefit of the technology itself.
Abe: So Caroline, go ahead, sorry.
Caroline: Abe, if I may just add on top of what they said. One specific AI machinery use cases is around power management. Intel CPU have multiple different power states which just lower the power of the CPU based on usage level. It's very different to something like [inaudible18:16] because it's so time sensitive and latency sensitive. So we are partnering with an operator like Rakuten to using AI and machine learning to automate adjusting the power level. So like Neil mentioned, green is the name of the game and we're all really focusing on conserving power usage. And that is a very good example of how do you use an AI and ML? Because of the fact that now you have this open ecosystem allowing you to really have a much finer control of the power usage. That's one of the best example I can think of.
Abe: So Patrick, back to you. Can you explain to us the significance of O/Cloud that being the cloud native deployment through this ORAN cloudification and orchestration platform?
Patrick: Certainly. We've seen telecom networks and cloud networks evolving almost in parallel to each other with different underlying technology over time. And 5G is really the first telecom network that is a cloud network in term of its fabric itself and in term of how it can behave. So what it means is that you can for the first time have a number of telecom function that behave like cloud function, and they can be deployed in the telecom network, or they can be deployed into the cloud, either a private cloud in the telecom network or a public cloud. What are the benefits of that? Well, the traditional benefits of the cloud, which are basically to provide an elastic on demand management of the functions, which means that you separate your hardware from the software, now your hardware becomes a lot less expensive than specialized hardware. And the software is basically provision on demand in order to satisfy your needs.
So, as we were mentioning, basically the network is going to follow the demand. And you all know that in traditional models, we were thinking about peak hours for our networks, and we have to provisioned our networks for those peak hours. Now, basically the network can adapt to the demand and whether people are working from home or working at the office, whether they're watching TV and consuming videos, 4k, 8k videos, or doing emails, the network actually adapt in real time to that demand and provide the necessary tools to cope with that demand in real time. So that has an impact on one hand on power consumption as was mentioned by Caroline, you have now the ability basically to scale down and to reduce the usage or rather the consumption of your network as the usage diminishes. So in the middle of the night, for instance, there's no point keeping all those computers up at max power when they are less used.
On the other hand, you now have the capacity if the telecom network is linked with a or several public clouds also to manage what can get processed where. And from a cloud perspective, the interest here is we talked about machine learning and artificial intelligence. We understand that artificial intelligence requires a huge amount of computing power in many cases in order to be able to identify patterns, in order to be able to create feedback loop into a new user adaptation that we mentioned and that usually is more economical to be done on a global cloud. So you now have networks that are able to do detection of those patterns, and then that feed those data to the cloud in order to do the processing and the machine learning itself so that we can all benefit basically from a network that is optimized. So our wholly owned subsidiary, Netcracker is one of the leader in the OSS, BSS and orchestration of cloud native solutions and is able to manage basically all those workload, whether on public clouds or private clouds, or both at the same time, in order to harmoniously enable use case such as slicing, such as dynamic traffic management across those different environments.
Abe: So Caroline, well, before I go to Caroline on testing and integration, Neil, any comment on O/Cloud?
Neil: Okay. Yeah. I mean, look, the part on cloud native is something that we've been pushing and literally at BT since before we understood what cloud native meant. So back in 2012 and the Olympics, we were trying to figure out how we could move capacity around from the different sports and that took us to NFV and then took us to cloud native. I give a great use case that we've got right now. We have a CDN on our network cloud, BT network cloud that's on the same platform our 5G core is, we're able to deliver beautiful content to our customers, video content that's tailored for their device via big screen, via handset, via tablet, via virtual reality glasses. We can actually optimize the content on demand and it costs us almost nothing to do it because we've put the two things together.
And then the cloud as a whole allows us to just manage hotspots much better. Allows us to do maintenance on a much more manageable way. Before we'd have an interruption, now we can change things in the network without any interruption, but I would caution one thing. There is a rush to the public cloud and the public cloud is pretty expensive. So, you know, we're looking at how do we really drive value? Because especially in a consumer market price and value is never more at the top of people's agenda as we come out of this pandemic. So, yeah, we love what the cloud players have built and the capabilities that they've got but they need to look at telecommunications through our lens and through the business that we deliver rather than looking it through as a big bunch of computer users that they can extract the most from because frankly our business model just won't allow for that. So, you know, new ways of working from the public cloud, guys would be really exciting so that we can really take those benefits into Open RAN, into distributed RAN in a big way.
Abe: So Caroline, off to you with the state of progress for testing and integration that I mentioned before. Really of these multi-vendor ORAN solutions and that through these end-to-end test specifications, can you comment on that?
Caroline: Yeah, [inaudible25:28] is the whole industry initiative to ensure that happens. You have ORAN Alliance announcing the open testing integration center. They've done [inaudible25:37] and I know Intel as a company has hosted [inaudible25:41] before the pandemic. I'm simply sitting on the board of [inaudible25:44] and we have labs for [inaudible25:47]. We have done [inaudible25:50] multiple times. We have announced a collaboration with ORAN Alliance to join and conduct [unclear25:57]. Then you also see individual operators, like NTT DoCoMo announced they have opened an ecosystem in Japan bringing multiple vendors together from the get-go to make sure that it's interoperable according to ORAN's spec. So generally I'm very optimistic as we are seeing the industry players all coming together and understanding the Open RAN really needs you. You are interoperable according to the open interface.
Abe: Any comment on testing and integration Patrick or Neil?
Patrick: Certainly. I mean, we also participate in all those forums. ORAN Alliance, [unclear26:40] as well as the NTT DoCoMo Alliance. We have a number of center of excellence that have been deployed in Japan, in India, UK, and soon to be announced in the US. So centers of excellence are done for pre-integration testing of Open RAN solutions with our partners. And we have taken quite a significant investment in [inaudible27:09] in our infrastructure and systems integration in the Open RAN space. And from our perspective, we are very pleased to that the ecosystem is really coming together. We are seeing multi-vendor environment, not only being tested and demonstrated, but actually being deployed commercially at scale. And that's really encouraging because it shows really that the real benefit of O-RAN is being realized, which is basically you can pick and choose vendors for every part of the given ecosystem, and it will work.
And right now at the beginning of ORAN, we were doing more integration work and now it's more interoperability work. And the difference really is that I think the vendors in the ecosystem are implementing the open interface, the open APIs in a responsible way. They're following the standards which means that basically you take a product from one vendor and a product from another vendor, and as long as they follow the standard, you can just guarantee that it's going to work. And then it's just a matter of organization for performance and security and other aspects, but the functional work of Open RAN is progressing quite well we think.
Abe: So, Neil, if you don't have any additional comment, I was going to go right into ORAN Security. Is that okay with you?
Neil: Yeah, I mean, I just want to add one little thing, Abe. I think there's a great amount going on in test and [unclear28:39] but where I think we've got a lot to learn, a huge amount to learn is actually running an ORAN network in the widest sense. You know, if you tweak one parameter on one vendor's device and tweak the same parameter on another vendor's device, they often don't do the same things. And I think there's a lot of experience, you know, whilst the pace of innovation has been fantastic we're still literally at what I would call the [inaudible29:11] of these networks in deployment. And I think we've got some, you know, challenging times ahead of us to upscale our people. And actually we're partnering with Intel on that right now to really build the right training and advance the courses that build that experience as quickly as we can. But there's still a hell of a lot to do.
Abe: Yeah. Let's talk briefly about security architectures and Neil I'll stay with you. So how is industry securing ORAN through the development of security architecture is really to enable 5G service providers to deploy and operate ORAN with really with confidence.
Neil: Yeah. I mean, Abe, honestly, this was probably the easiest part of ORAN in my mind, because when we designed 5G, we designed it with security from the heart. You know, if you look back at GSM, 2G, 3G, 4G, we weren't really thinking about security in the way that we think about it now. And we're using industrial standards, you know, military grade security, you know splitting the control plane and on the data plane so that it's harder to do, attacks, encrypting everything. Again, thinking about privacy from the start and using, you know great proven track security platforms, you know, even IPv6 is the basis for 5G. So in my mind, this is the simplest part of 5G because the whole way we designed it was with security at the heart. So when an Open RAN vendor comes along, it's really easy for them to build their solutions using the encryption, using the security methods, using the authentication and the control plane methods to keep it really secure and to keep our users privacy at the top of the agenda as well. So, you know, that, I think the work that we did in 5G has made ORAN possible because I think if we hadn't done that, you know, trying to do this with 4G would have been a lot harder.
Caroline: I couldn't agree more with when you said the fact that you have open standard is of the best way to ensure security. It is a light bulb, so you could see the interface. And also that it's based on the same CPU that we supply to the enterprise and the cloud, including the military space, right and you mentioned. There's a lot of security put in place at the CPU level, securable, secure device onboarding and so on. And that now being ever able to supply as part of the foundation for Open RAN in 5G. It really sets us well, make sure that we have [unclear31:53] security.
Abe: Patrick, secure networks for ORAN, any comment?
Patrick: NEC takes security very seriously. We're one of the world leader in biometrics and that is being applied for a number of security application for a number of governments. So we have implemented from the very beginning zero trust framework for our Open RAN products, which allows basically to have strong authentication, strong algorithmic encryption between inter-layers and every interface. And that allows basically to have those complex systems in a multi-vendor environment to be able to be extremely secure because you cannot just introduce a new element in those systems without it being authenticated and being able to receive and manage the encryption [unclear32:58]. So as Neil mentioned, 5G as a technology generally speaking has been designed with security from the outset and Open RAN from that perspective because basically it is mostly a software. Cloud native is actually able to add an extra layer of security by allowing this zero trust framework to be implemented.
Neil: And Abe I think the other thing is AI and anomaly detection gives us another layer of security across, you know, the end-to-end networking architecture. If you think about anomalies are things that have gone wrong. Anomaly detection which we can build into the heart of this platform because of the openness really allows us to take that security even one step farther than we've ever been able to do before. And I think you know, that's something we're very excited about and our customers are really, really excited about.
Caroline: Well said.
Abe: Really well. That's a great place to stop. I mean, you know, the discussion around ORAN and today's Open and Smart RAN, it's important to kind of stay ahead of that discussion. So that's what we try to do. And by the way, I want to just recognize the ORAN Alliance. We do follow their white papers and such, and that gives us some good content to discuss in this format as well. So I wanted to recognize that. But to have all you three talking about this topic and again, this steady drum beat of staying ahead of the discussion is really important, so we appreciate everybody's time. Caroline Chan not only appreciate your time, but appreciate Intel making this possible today. So thank you for that.
Caroline: You're welcome.
Abe: Thanks. And Patrick and Neil, of course you know, your time is valuable and we appreciate you being with us this morning. And we hope you will come back. Hopefully, we see all of you, by the way, in Los Angeles at the end of the year. At this point, it's kind of a tossup, but we'll see what happens, but hopefully we can see each other in person.
Neil: We're trying.
Patrick: Yeah. We certainly hope to.
Abe: Absolutely. And once again, for all the viewers out there, thank you for watching Open and Smart RAN, The Deployment Scenarios for this executive session on demand that on August 12th, please visit thenetworkmediagroup.com. So long.
For any inquiries, please email anejad@thenetworkmediagroup.com