Munish Chhabra, Radisys:
Good afternoon everyone. I know this is post-lunch session, but we'll keep it interactive, so keep everyone awake. Yeah. Okay. Welcome everyone here. I'm joined by three speakers. We have Jim Estes, CEO, Dense Air. He has 35 plus years of experience in multiple roles in telecommunication industries industry, right from building products to deploying product now and changing the word from how the infrastructure is deployed. Today we have Vinay Ravuri, founder and CEO of EdgeQ Inc. Again, Vinay has played critical roles in companies like Qualcomm and is now changing the way SOC is looked at it and also joined by Arun Bhikshesvaran, CEO of Radisys and he's responsible for executing Radisys strategy in terms of transforming the telecom industry. Again, Arun has played multiple roles previously with Ericsson to welcome everyone. We'll go over a few important questions, which I think everybody has their mind and then we'll open up the floor for questions for everybody. Alright, so starting with, I'm sure this is everybody's thinking where 5G is expected to play critical role from evolutionary and revolutionary capabilities in wireless networks. So we look at how this panel symbolizes with the revolutionary nature of 5G networks. Starting with this, I would say Jim, maybe if you want to start with how does your approach to bridging the infrastructure with cloud networks changes the traditional network deployments?

Jim Estes, Dense Air:
Thank you. Well, the great thing about this architecture with open Ran is that we're able to lever a lot of really talented partners and bring it all together in a pretty amazing 5G solution. We really have looked at this as a way I view it as, I call it Tower 2.0. You think of the tower model in the early days was very, no one wanted to do it and now it's commonplace and we're now at the next inflection point, which is really to share the active infrastructure and this is what the architecture enables. We can actually have multiple operators operating on the same radios and sharing that to get the best efficiency and economies of scale like in the tower model, except we're doing it with the base stations

Munish Chhabra, Radisys:
Thank you Jim. And Arun, maybe if you want to share your views, especially with your experience with Ericsson, how is 5G changing the way how the networks deployed today versus traditional networks?

Arun Bhikshesvaran, Radisys:
Yeah, I think that's a good question. Munish, like we were talking about a little bit earlier this morning, every generation of wireless has been deployed basically for the same use case, which is people talking or watching YouTube videos maybe, right? And I think what we want see in what the industry wants to see in 5G is a very different way of monetization, but also very different way of serving multiple use cases. At the end of the day, not everybody is driving a German car and all cities are not consisting only of highways. There are speed and B lanes. And so you've got to have diversity in the way you do things, and that is the beauty of the partnership that we have on stage. You've got a new silicon vendor, a new way of deploying networks that makes it accessible to multiple people and the role of software coming in and everything working together underpinned by the cloud and that is very, very different from what any other generation has achieved in the past. I think that's the beauty of the, whether you want to add something.

Vinay Ravuri, EdgeQ Inc:
You said it really well. 5G is very different from 4G, I call the evolutionary part of 5G is like 4.9 G, more bandwidth and all that. But the revolutionary part is to enable new use cases beyond the phone. Phone is one modality. You have cars as Aaron just mentioned, and there are more to come and this is also a step towards six G that's also there. So I do think that even though the revolutionary part of 5G is not seen today, but it is an evolution of these things, it takes time, but it's just going to be the non phone that enables these use cases.

Munish Chhabra, Radisys:
Thank you gentlemen. I think this was very insightful. Let's go over to the next topic, which is about the scalability of the solution that we are talking about here, especially with respect to using cloud native environment and in this particular case they're using GCP. So maybe Arun, if you want to start with how Radisys is enabling cloud infrastructure through 5G ran application.

Arun Bhikshesvaran, Radisys:
So absolutely. I see the basic premise of being able to run ran software in the cloud is something that we are giving true meaning to. Other people talk about doing it, but they're mostly doing it in private clouds inside telco environments, what we are bringing together is a true public cloud deployment and that is massive. And you talked about GCP, there is also a component of GDC on the distributed cloud, right? So when you start to put the DU closer to in the GDC to close up deployment and having the CU run in the central cloud more from a GCP perspective, that simple capability sounds simple, but you got to open up the interface and make sure everything works in a very structured manner. And that is only the first step because then after that you can separate the CU into a control on a user plane and start to build cardinality in a very different manner.

And that is how you start supporting multiple cells and then you get auto scaling and everything. Now if you are a traditional OEM, you have to think about feature parody and how you take your current code base and you start to modernize it. But because we start from a different level, all this comes to us natively, we think about it and start to go implement it without any baggage behind us. I think that is a huge plus and I think it's great that Jim and Dense Air is going to take it out into the market and create a showcase that other people will actually learn from.

Munish Chhabra, Radisys:
Exactly. So my next question, Jim, is for you that how are using GCP helps both in terms of scaling but still ensuring the security and reliability in the shared network infrastructure?

Jim Estes, Dense Air:
Thank you. Just to build on what Arun said, the great thing about working with GCP is I as operator of this network don't need to be in the server business. I don't want to own a bunch of bare metal servers and have to maintain them and all the transport and associated costs. So GCP has a global reach, we're levering that we can take advantage of their scale and all the security features that they've implemented in the cloud. But then now I just have to basically attach the radio to the cloud and so the DU and the CU are running the cloud and ability to deploy rapidly is massively extended beyond having to deploy bare metal server. So it's a very exciting opportunity.

Munish Chhabra, Radisys:
And maybe Vinay if you can also talk about how does the cloud environment or cloud native applications throw challenges on the SOC?

Vinay Ravuri, EdgeQ Inc:
Yeah, there are many. Well first thing is just to reiterate what Arun and Jim just said. The benefits are that economies of scale that you and the know-how that comes from other applications can be adopted. So hence it's a more efficient network. A lot of the software could be reused. So there's a lot of value in not reinventing the wheel all over again, which cellular industry has done too many times. So this is a good way to go. Now everything eventually has to run on hardware and chips. So when you run something in the cloud, you try to make it as general purpose as possible because you want to run different applications, not just one thing but many things. So that creates an interesting challenge for a silicon vendor like ours because on one end you want to be very cost efficient, very power efficient.

On the other end you want to be general purpose. These are vectors that pull against each other. So that's the challenge and that's why I exist as a company because that's also the opportunity and traditionally this is not how things have been done. So this is the way HQ was started is that how do we run cloud native applications or take things like containerization Linux, which is a very common thing. How do you run it efficiently in a modem or in a RU or a du? And that's the model that we built. If you look at the traditional approaches, that's not how they've done and that's also the challenge for us, but there's many more coming because cloud continues to evolve, which means chips that go in it have to evolve as well.

Munish Chhabra, Radisys:
So continuing to actually on that topic, especially the advancement which is happening in the 5G area. From SOC point of view, how is your approach of base station on a chip changes the evolution for the current solution and especially help dense air in the solution which is being discussed here?

Vinay Ravuri, EdgeQ Inc:
Dense Air should say what they like, but we are thankful for the partnership with Sair and Redis. So the thing that our customers see in HQ is that we kind of did a clean slate approach. We didn't do this based on our previous generation because we didn't have a previous generation first thing. Second thing is we looked at the industry and said, okay, what is the best of breed that I can adopt rather than reinvent? Okay, so further the modem or the communication part, we took risk five. Instead of reinventing that world ourselves, which a lot of modem companies do, we decided not to do that. So there's some goodness that comes with it, but at the same time, there's software that's out there that has to run on top of our silicon, which needs ecosystem to be there and things like risk five doesn't really have a great ecosystem, so we put arm on top of that modem, which is inside the ship.

So these are decisions we had to make. Now in 2018, this looked really strange to everyone like mixing two different technologies and this looks more like a computer than it looks like a modem today, five years later people say, oh, well that was such a good idea. It's like you are far ahead. Honestly, it was an obvious idea. The industry just needs to adopt what's there.

Munish Chhabra, Radisys:
So Jim, to you, how in terms of using HQ as a SOC of choice helps you in terms of developing the solution faster and especially as part of your small cell?

Jim Estes, Dense Air:
Yeah, thank you. Just to build on what Vinay said, one of the big things is a significant reduction in cost. So I can build all of this on one chip rather than have as many as four or five more chips to do the same thing. The second thing is is that I have a radio that I can do multiple bands and functions all on the same one. So I can do CBRS, I can do C-band and I can do band 41 all on the same chip set. So now I can provide a R ru both indoor and outdoor. This supports both licensed and unlicensed spectrum all in one radio. It's a pretty dramatic savings because we can share the fronthaul on that as well.

Munish Chhabra, Radisys:
Thank you. Arun, back to you. How does it impact the RAN implementation or architecture given that you have more capacity available right on the SOC?

Arun Bhikshesvaran, Radisys:
Yeah, I was just going to talk about that, Jim. We went and met with Vinay's engineering team at Bangalore over a weekend and on a Saturday morning they opened up the office for us and suddenly, I don't think they were expecting this big crew coming from Radisys and when we told people that we were going to go meet with EdgeQ, there were more people who wanted to join us. So we all went into this conference room and we started asking all these questions as to how does this really work, blah, blah blah. And then your head of engineering Hari says, look, hang on for a second, and he brings out a tray that has just come from the factory with all the chip sets and he says, look, this is what it is.

This is real. It is dramatic. When you think about what a base station on a chip actually looks like and what you said vinne about having the arm cores and the disc five architecture running our software with all the experience that we have, compressing it, running in four cores, et cetera, you put everything on one deck. The cost reduction that you get in terms of having a highly efficient design, this is revolutionary for us because we know how to run it on 20 cores and 32 cores, but when you start to say, Hey, everything that you have designed is here in one chip, I think that is groundbreaking. It's a new frontier for us.

Munish Chhabra, Radisys:
Absolutely. I think this not only impacts in general the densair solution but also the industry overall in terms of bringing the cause down, especially for small cell. So coming back to Vinay in terms of the current solution that we are discussing, I'm sure and this audience would have questions about split six versus split 7.2. So why split six is more effective here compared to 7.2, which is more popular?

Vinay Ravuri, EdgeQ Inc:
In open ran? For those that don't know, there are different splits that you can split a base station 7.2 implements the physical layer in the RU and then the upper PI is done in a DU as it's called, and in split six with the entire PI is implemented in one place and then you aggregate a bunch of these things into a single deal. So architecturally and dense air does that, and our silicon is flexible. It could be cut in any way, but it does have a lot of advantages when you do the split six this way because the fronthaul bandwidth gets greatly reduced when you process the whole file. What people don't understand is when you are, let's say downloading something, let's say you're downloading like 100 megabits per second, hypothetically the bits that go in the air is not a hundred megabits per second.

It's almost like base 700 megabits to a gigabit per second if it include the number of antennas and include all these things. But so when you have fronthaul where you have to transport these bits, it becomes expensive. So that's like almost a tax that you have to pay. So doing more of the radio processing if you will, or the base band as it's called upfront, it essentially reduces the cost of deployment but also architecturally opens up new application.

Munish Chhabra, Radisys:
Yeah, Jim, so how has this helped see from a solution point of view and what were the choices that you had to make between the two?

Jim Estes, DenseAir:
Thank you. Split six is a big opportunity as an operator of a network. So we denair build, owns, operates these networks as a neutral host and we then provide capacity as they need it to different operators both on the enterprise side and on the carrier side with split six, I don't need to use dark fiber out to all of it, so that saves a lot of money. Also, the latency of tolerance is much more flexible, so that allows me to actually run longer runs without having to optimize a network around latency in indoor. The interesting thing about that is you can many times use existing IT cabling and be able to do the front haul. Then split six allows you to do that. If you were doing 7.2 you would have to haul fiber everywhere or so it's a big cost savings really drives down our unit cost. That's great,

Munish Chhabra, Radisys:
Thank you. Jim. Arun, back to you as Radisys is, you have seen both 7.2 evolution work with so many different radios on fronthaul and now split six options with small sales. Which do you think suits better from Radius's architecture or even for this particular use case around cloud?

Arun Bhikshesvaran, Radisys (17:18):
Yeah, so I think the beauty of split six and what we do with EdgeQ is that you have a specialist in the file layer that you completely rely on to take advantage of. So you let Vinay and his team excel in that domain and we excel in software and we run the software along with his core complex and that brings together an ecosystem of people who are specialists in their domain. That is the beauty of split six and what we do. But you mentioned something earlier Munish about this being a great option for small cells and I have discussed that this could also be a very good option for 7.2 because what you then start to do is to have a single card implementation on which you're running your DU and CU and you can have a very efficient, power efficient, cost efficient infrastructure that runs at the bottom of the tower and then comes our flexibility that we have interrupted with a number of 7.2 radios and this can start to bring a very different perspective on where HQ technology along with our technology can be deployed.

This is not only for small and deployment, right? As we scale and we go to other kinds of shared infrastructure, macro network deployment, it starts to become a very efficient use case and if I can do a small pitch for vinne, the ability to support both 4G and 5G, and this is Jim, we have talked about this as well on a single SOC also starts to become very, very important as we talk about the evolution of networks. Even here in Europe for example, people are looking to have at least 4G as a fallback to go in and say that you can support both technologies, same infrastructure, this is what we talked about yesterday. You can start with a 4G if you want. You can upgrade to a 5G and over a period of time you can tune down the 4G for example. Tremendous flexibility that you get and for rapid replay kind scenarios, this becomes a very, very efficient way to look at it. So there is a lot of flexibility that we get with these solutions and then getting Jim and his team giving this flexibility to get started in the US I think is such a great opportunity. Good.

Vinay Ravuri, EdgeQ Inc:
I don't need any more sales guys. He is done for me just to extend what Arun just said, but thank you for those kind words by the way. We were asking about cloud and how this can be used, reused, adopted, if you look at Google or any of these guys, they don't have different hardware for Gmail versus YouTube versus it's the same hardware and not even, it's not, it's just the same hardware. They also the stack, the software elements, the foundation elements are actually the same regardless of the application. So architecturally setting up a platform silicon as well as some layers of the software to be common across different implementation of the 5G RAN is key because if you look at historically there are not that many base station companies because it takes forever to build one of those things and it's like you need a certain type of mindset, you need the expertise, forget all that. If you build this in the right way, then you reuse this platform and that was the goal of the company as Arun articulated and that's what we want to see more and more.

Munish Chhabra, Radisys:
Actually leading on to that, I would say maybe Jim if you want to share that the ecosystem that is created with Dense Air radisys edgeq, how is it helping reducing the cost and increase the flexibility for deployments across enterprise and network as well?

Jim Estes, Dense Air:
We see dramatic cost reductions and look at the open ran architecture because we can use software that's been developed in other places without having to develop multiple times big cost savings. The other area that we see a lot of cost savings, you talked about IT world and the enterprise world, we're bridging the gap between the enterprise world and the mobile network world. I mean mobile networks were always this thing that was completely unique and then everybody, oh, I don't want to work with the carrier or whatever. We go down this path and we can actually make this look much more like an IT implementation for indoors for example, especially if you're running split six inside, I don't need to have a bunch of you think about a das system and you're pulling coax all over fiber. We don't have to do that. One of the things we found is that it scales down much better than DAFs does.

So if you think about a smaller facility, let's say 250, 300 50,000 square foot office building das is very expensive to do. So that's problem number one. We don't have that scale down issue using this architecture. The second thing is is that I don't need to bring a signal source to the site everyone. So everyone knows this is a challenge. You need to get the m and o to bring a signal source. They may not think it's important, they don't want to do it, so they're not on your network Here we integrate the neutral host to each of the operator's core networks and now I don't have to do that every time I do it once and I can reuse it everywhere. So that's pretty significant savings.

Munish Chhabra, Radisys:
Thank you. Jim. Maybe Arun, don't know if you want to build on top of it. How does this ecosystem helps in terms of faster deployments and even again emphasizing upon reduction of cost per deployments of networks?

Arun Bhikshesvaran, Radisys:
See one of the biggest thing is when Jim goes and deploys in special venues, they don't want to have multiple deployments from every single operator in the country. They want to deploy it once. They want to have it flexible and they want that to be good. So it has to cover every single part of the environment. We've talked about some use cases where it's not just about indoor but it's about also covering the parameter right outside the loading dock for example. So you want coverage everywhere and you want it to be as seamless as possible. So for every operator to go in and deploy that infrastructure, it's just a waste of money.

You don't need it. And having the new tool host concept come in there, make it available and then taking advantage of the underlying cloud environment like we have discussed, it really gets the distribution there with great penetration. So the calculation on cost savings is obvious just from pure sharing, but I think the calculation from cost savings in terms of modern technology just accelerates that calculation.

Munish Chhabra, Radisys:
Vinay coming to you now, apart from giving run for money to the big companies and reducing the cost through your innovative submission, what flexibility do you provide as part of this ecosystem or what flexibility does the ecosystem provides for faster, quicker deployments?

Vinay Ravuri, EdgeQ Inc:
Yeah, that's a good question. So other than the fact that we integrate a lot of things and gives you low power and all the flexibility of the chip that it has, what we have done or recognized was that it's not enough to just build a chip and say you write your software, so we build a layer one FI that is hardened ready to go. So you don't have to work on that yourself.

One of the bigger companies does do that, most others don't. Then on top of it, we recognize that to kind of accelerate these things, so we're a chip company, we don't actually have to do layer two, layer three that could come somewhere else, but we licensed your stack early on Redis stack and we worked on showing that it can be integrated and you can get very high performance because otherwise people see a slide and you can is something that's like, okay, maybe I can, maybe I can't. So getting that demoed and show that this can be done and show proof makes and I don't want to go build a layer two, layer three stack, that's too much work. We're a chip company I want to say as a chip company. So ecosystem there is necessary. So those are good things, but it's also there's a lot of other things that have to come together over time. It's not as simple. If you take wifi, I like to use that example like a company like Netgear, they can take a chip, put it into production in six months to nine months, you can get a wifi router at home. It takes you 10 minutes to configure that afterwards and you're done. You probably have to add a few zeros next to that for 5G. Okay, so the simplification is that the ecosystem has come together well for wifi.

5G can also do that. It's just to take time and we have to have enough players, more players like you, more players like you that bring this to some simplicity.

Munish Chhabra, Radisys:
Absolutely, and I think as you simplify things, this also throws other challenges. When you create open and flexible architecture implementations, this will throw many other challenges. So maybe we go over the panel in terms of what challenges do you guys face while creating such networks?

Jim Estes, Dense Air:
So this architecture and this approach allows us to do things we never could do otherwise. So if I could buy what I just described, I would buy it, but it doesn't exist. In fact with the 4G and 5G, we had one major mobile network operator in the US called the unicorn and we're going to build it and we can only do that because of this ecosystem. But then because we have all these different elements, things like how do we make sure we have interoperability and achieve the right kind of quality of service for example. So there's a lot of really interesting things in the cloud with GCP that they do on demand scaling instead of doing traditional QLS where you tag the payload and give it though this is more important than that. What they do is actually scale the capacity up and down in the cloud as you need it. Pretty revolutionary. They do it in an IT world every day you don't even think about it. It's not been really done on this part of the world. So that's really good from a security standpoint. That's the other area we have to spend a lot of time looking at. We need to go through every element in here, make sure that it meets all the security requirements. We don't want to, it creates issues with that, right? And that's a key element of this whole architecture is making sure that we have adequate security.

Munish Chhabra, Radisys:
Maybe Vinay, if you want to share your thoughts around especially the challenges by creating these networks because it's not easy either and especially the industry to accept it because it's very different the way it is done now.

Vinay Ravuri, EdgeQ Inc:
Yeah, one of the biggest things is why has an OAN really taken off in a big way is the ecosystem challenges exist. It's one thing to say we disaggregate and you got this piece, but then in integrating them and making them work in a seamless manner because that's what people buy. They don't buy components, they don't buy segments of things, they buy a solution. When I am on my phone, I don't want to think about how this thing is connecting, I just want to make use of it. So the challenges, there are many, like I said, the challenges for wifi existed a long time ago, but they've worked through it and it's seamless.

That still exists with 5G. It starts from how do you even benchmark a O RAN system is a question like we take system one versus system two as system three. How do I compare these things? What are the metrics like machine learning there there's something called ML per, there's no equivalent to that here. And who is the systems integrator who integrates them? What's the one 800 number to call? So there are many of these challenges that have to be addressed. Cloud native is a good way to do it because then it just goes in the cloud. Nobody knows what happens but it works. So therefore that's it. It's a good way to address that because then you're not reinventing that wheel, but it's going to take time for orran to be the market that a non orran market is today for it to be fully covered.

It'll take time because the ecosystem has to get formed around it.

Munish Chhabra, Radisys:
Thank you Vinay.

Arun Bhikshesvaran, Radisys:
See, one of the important things, Vinay and I can create technology. At the end of the day, you need a buyer and you need a buyer who is willing to take it, accept it for what it is, work with it to mature it and deploy it. And I think that's the beauty that we have in this. The biggest challenge, the biggest help we have received is from Dset and also the company that backs you driven by a philosophy that comes from SIP in terms of look, you got to get wireless coverage out there for the people who really need it. And there is a big movement behind that and modern use cases connecting to cav new, et cetera, that driving force is needed. Very often what you see is that the traditional operator community, even though they talk about monetization, the ability to go in and do these things, take action, put money behind the cost that is lacking because they're too wound up in their classical way of thinking. I think that is probably the biggest challenge in commercialization of any technology and I think Jim and the backing of sip, that's a huge thing that we have in the ecosystem.

Jim Estes, Dense Air:
And there's one other ecosystem partner beyond the one you talked about is we have a world-class ODM partner in WNC and Vinay introduced us to 'em mouth them was sparring with their CEO. They are first class and we would not be able to get the time of market that we're doing here. We don't have to go build a factory and they're going to build us in Taiwan for us. But the whole thing is coming together to deliver what we need.

Munish Chhabra, Radisys:
Thank you Jim. I think that was super helpful, but Arun, back to you. You avoided my technology giving me all my, yeah, so would still want to all otherwise I still practically on all this goes would still like to hear from you the technology challenge of being first in the industry to create open and flexible networks. So we should also talk about some challenges with respect to what kind of challenges does it impose in software? Because software in general people think about is easy, right? You can do anything in software, but the reality when you have to create such open networks, what are the challenges that you face?

Arun Bhikshesvaran, Radisys:
See, one of the biggest challenges I think we have faced, which is also the advantage here is when you go talk to a traditional mobile operator for them to consider you as a software vendor, the first thing that they talk about is feature parody. And it's not featured parody for today's technology. It's featured parody for three generations of technologies and common sense the calculation simply doesn't work. Why would I go create software for two G technology? But if you put that as an entry barrier, it creates tremendous complications because now I have to figure out not only how to create modern software with DU and microservices and make it cloud native and everything, but I have to deal with the non-standardized, non-open version of monolith for the earlier generations and I got to make these software coexist. That is a very, very big complication for us. I think in this particular case, the big advantage that we have, if we focus in on something that is very sharp new, that allows us to start with a clean slate approach similar to what then I did without a lot of the baggage and then we focus on, hey, how do we make this cloud native?

So the challenges that have been posed on us in the case of D, not so much because we don't have that big threshold to cross that allows us to come in with a clean implementation of 5G SA and start from there. And for that we come in with software that has been designed and tested for that environment, interrupt tested with top tier OEMs, et cetera. That product grade that you get from the the door, that is a big advantage, right? So I think for any new technology provider, definitely for us from a 5G software perspective, the challenge to address legacy technologies that was kind of removed in this case and now we might do something on 4G, but that is still much better than having to go all the way back to two G.

Munish Chhabra, Radisys:
Perfect. I think these parts were very helpful and insightful. So I'll open up the floor for audience for any questions. Let's ask some difficult questions guys.

Audience:
If you would describe the perfect customer for you because you need to enter with this idea also to the big operators that are traditional thinking and this is so new. Would you mind to describe what is your first priority?

Jim Estes, Dense Air:
So there's actually two ways to look at that. We talked about enterprise and the traditional m and o environment. Initially we're seeing very large demand on the enterprise side. So many of the enterprises go to the MNOs and they say, your coverage is horrible in my building, can you help me? And they're like, well no, we don't have the money or we're busy or whatever. So there's a venue pays model that's developing out there. It's similar to what happened in the UK with Jots originally where they are basically saying, we will allow you to rebroadcast our PLMN and integrate your neutral host into our core. But then you go to the venue and the venue pays for the deployment. So that's one model they're very enthusiastic about that they don't have the money to spend on that. Beyond that, it really is now for let's say outdoor small cells for capacity and for coverage infill.

Now we're trying to help them solve a problem that they may or may not have the capital to do that we bring a couple things. We do build, own, operate and maintain. So we actually will pay for the CapEx, build it, and then we have our own knock and we operate it and that integrates into their network and it looks like any BAU kind of base station from day one. So they can hand in hand off all the parameters that you would normally do as if it was a base station that you purchased from someone is there, but it's a virtual base station effectively and you're only paying for the part you use. I was quite skeptical that people would be interested in that when I first met the investors because I mean I thought are ever going to get these guys to ever even consider it.

And it's interesting at a relatively high level in the organizations, they're very interested in this, but it's then how do you triple that down into the organization within the existing ran groups and things and make it part of their new business as usual. We're actually seeing pretty good traction on that, but it really is starting more indoor enterprise. And you mentioned the, we call it near outdoor. So if you think about do the indoor, I'm going to cover the parking lot, it's a hotel. Let's say I'll cover the pool for the beach area and all that. So the venues are very willing to pay for that because it's a big part of their customer satisfaction and it really doesn't cost anything to the MNOs. They go put us through validation like they would anything to make sure that it meets all of their requirements that isn't going to cause problems with their macro network and then we can broadcast their PL nines and that solves everybody's problem. So little lengthy there, but it it's kind a two phase or two-prong approach.

Munish Chhabra, Radisys:
Thank you Jim. And do you see Jim, just continuing on that, do you see the industry or the tier one operators accepting the fact, especially across shared infrastructure? I'm sure the audience also has this in their mind.

Jim Estes, Dense Air:
As I said, I never thought that they would, but a lot of it comes down to just economics and what they have the bandwidth to do. So the fact that we can come in and help them, there are things we could do now that they may not be able to do for a couple of years. And you look at 5G, they're rolling a lot of this out on the macro network, but they may not get to a lot of the small cells for a long time. So if they wanted to accelerate part of that, they're actually pretty open-minded. Again, as long as it meets all the requirements that it integrates into their network as any other base station would, it's not causing problems of their macro network integrates to their o and m as well so that when there's an alarm, it shows up in their knock and ours as well. They see it and they know it's there, they can hand any hand off. So it's evolving. There's definitely interest there.

Munish Chhabra, Radisys:
Thank you, Jim. Any other questions? Anybody?

Audience:
Yeah, thank you everyone for the insightful thoughts Vinay I have one question for you. How is AI going to change the layer one posted on SOC? Your thoughts on that? Thanks.

Vinay Ravuri, EdgeQ Inc:
Yeah. So who is my perfect customer? The one that pays full price and doesn't ask any questions? We look for scale, but coming to ai, my investors said if I say AI once a day, our valuation will go up. So thank you for that question. So in my previous job I was at Qualcomm and I worked a bit on the AI part of it. Machine learning has lots of potential to change the RAN and software aspects of it. So I look at AI as there are two areas in the ran. One is how to make 5G or 4G better using ai, meaning implementing the physical layer or layer two, layer three, but use machine learning algorithm instead of the traditional approach like a scheduler that schedules users learn a pattern and optimize that at the physical layer.

Understand the spectrum when you're in low SNR regions, like where signal is not good if employing traditional algorithms give you very predictable results, but in those areas you want to find something different, a much more higher performance results, meaning a single bar on your phone, but you still want to get a very high throughput or medium throughput in that. So machine learning actually does a better job there compared to traditional. So we implement things like that. Now the second bucket, which hasn't yet taken off is moving machine learning models to the edge. A large portion is still in the cloud, whether it runs on let's say the end device or the closest to the end device being a CU DU the RAN in general. There are a lot of discussions and work being done here, but it hasn't come through yet. Take as an example, you have a bunch of cameras that there's live feed, you don't want to store things that are not important, but yet you also don't want to send all this to the cloud to figure out what's important, which means ML is a good way to summarize these things.

But you also don't want that in every camera because you want to amortize that or aggregate that rather into a central place, a semi central place that's a base station, that's an area for example. So there are many applications like that today. The problem statement is not can machine learning help? Actually it can. That's not even a problem. The problem statement is how do you implement them in a cost-effective manner and is there an ecosystem to make use of that so that not everybody has to do it themselves in the cloud. There is AWS has one, Google has one, right? The RAN doesn't yet.

Munish Chhabra, Radisys:
I think this was very interesting. Anyone else, any in questions?

Audience:
A couple of quick questions. Question number one, Jim is how does this solution sort of standard compare if you were to take the current path with advanced distributed antenna systems? It's ubiquitous, right? Do you see any challenges trying to break that strangle hold?

Jim Estes, Dense Air:
So you're talking about das for indoor venues, right? Yes. So we have seen tremendous demand at what I'll call at a small to medium size enterprises. So if you had a building that's, like I said before, I think let's call it two hundred fifty, three hundred 50,000 square feet down to let's say 50,000 square feet. das does not scale well at that level. And even if you installed it, the odds of them ringing a signal source from each of the MNOs is probably not very good even if you paid for the signal source. So because we use the neutral host integrated into each of the MNOs core networks, they don't need to bring the signal source to the venue. So that's a big thing, but not only in terms of cost, but in terms of accelerating the deployment of it. And then in terms of the, because we're again standard IT cabling and things versus coax and fiber like you see in a DA system that's less expensive as well. And you don't have to amortize the cost of a base station hotel across that particular venue. So pretty dramatic savings. I mean we're seeing in some cases 60 to 70% cost reduction over das at the smaller end. That doesn't mean you can't do a million square foot venue, but that's in that kind of lower end is very, very high demand and at a very significant cost savings over das.

Munish Chhabra, Radisys:
Okay, thank you. Thanks Jim. Anybody else? If not, then I think we can wrap up the session today. Thank you for the panel for sharing insightful thoughts around how 5G can change the world and virtually the networks the way they're deployed today. And I believe next year we'll see the network deployments right? So around this and learn about the challenges that we have faced while deploying these. And thank you for the audience for asking interesting questions. Thank you everyone.