Guy Daniels, TelecomTV (00:24):
Hello, you are watching the Green Network Summit, part of our year round DSP Leaders coverage and it's time now for our live Q and a show. I'm Guy Daniels and this is the second of two q and a shows and it's your final chance to ask questions on energy efficiency and innovation within the network. As part of today's summit, my colleague Ray Le Maistre hosted a panel discussion that investigated the benefits of building a Green RAN. And if you miss the panel, don't worry because we will rebroadcast it straight after this live q and a program or you can watch it anytime on demand. We have already received a number of questions from you. In fact, we received about a dozen towards the end of yesterday's q and a, so we will be answering some of those first, but there is still time for you to send in a question.

(01:21):
Don't wait too long though because we only have about 45 or 50 minutes today. Well, I'm pleased to say that joining me live on the program today are Joan Triay, who is Deputy Director and Network Architect at Docomo Euro Labs and Rapporteur for the ETSI ISG NFV; Sarat Puthenpura, who is Chief Architect Open Radio Access Network, ONF and the Aether Project at Linux Foundation. And Francis Haysom, Principal Analyst with Appledore Research. Hello everyone. Good to see you all. Thanks so much for taking part in the program today. Let's get straight to our first audience question. We have got a lot of them and let read the question to you. With the transition to software defined networks and cloud native operators, what standards are emerging for understanding network function and service level energy consumption of software development and operation? Great. First question, start with Joan. Are we able to come across to you for your comments first?

Joan Triay, DOCOMO Euro-Labs & ETSI (02:33):
Yeah. Great, thank you. Yeah, I would say that one of the to consider is that even if the topic is about the software defined networks, we should just not only talk about the software itself because we also need to consider where this software is actually being executed. So when it comes to standards that are going to help us determine whether we are going on the right path to, for example, save on energy consumption, we need to discern the inputs that we are going to get, not just only from the software itself that is being run, but also of the infrastructure is being that is being used in this aspect. I think that we need to then probably categorize what are the standards that are usable into two categories. One is the category I would say that are the standards that are related to telemetry or measurements or KPIs and other standards or best practices are really more about how the software is being built When it comes to standards for the telemetry, as I said, I believe that we should not only restrict to the aspect of the software, we also need to build up on the stack.

(04:02):
So we need to build up from the bottom because it's where actually the N energy consumption is happening and in this domain in terms of infrastructure, the standards that are being delivered by the ET C, like the European norms coming from the environmental engineering and other technical bodies are quite essential since they determine the actual KPIs and measurements and even the best practices for collecting and taking the measurements themselves on physical devices including telecom equipment. Very similar effort is also being done by the ITUT and there are several recommendations about that. And if we start them building up on top of this stack of the fixed critical infrastructure, then what I would emphasize are, for example, standards coming from organizations like the H-C-I-H-G-N-F-E that they have defined, for instance, measurements that are related to the containerized workloads and virtualized workloads that are being used to actually deploy the software entities building the network.

(05:16):
So they have defined measurements taking into consideration also the assumption that there are certain capabilities in the infrastructure in the platform like tools from KPL and kandra that are able to actually either collect specific information from the CPUs on the energy consumption or estimated using some form of machine learning. Now those measurements are quite important because then we can actually relate them to the software entities that we deployed, that those are the network functions themselves, those are specifications as I mentioned, defining the bottom line of these measurements for network functions which are then consequently also being used by other organizations like for example, 3G BP in defining their measurements that are associated to the actual network functions that concern to network functions that build the actual network like the 3G BP mobile network. Now this is I would say the group of the that relate more to the telemetry When it comes more to the software itself, my understanding is that there are other standards like for example, the ones defined by the Green Software Foundation that some of them I believe are already iOS ISO standards.

(06:41):
Those are standards. Basically what they define is actual KPIs to determine the impact on the energy consumption and the carbon footprint of the applications that are being deployed. And that gives also very good guidance to actual developers on how to target, what to optimize or not in the building of their applications. And really to this aspect is also about practices and principles that are also being described on how to improve the actual software development, like how to try to minimize the number of computer cycles that are being needed from the infrastructure to actually execute the workloads. So as I said, I would say that we need to tackle or we need to be aware that the standards are not just only one, there are multiple of them for the benefit of the operators and the industry. I believe that a combination of them need to be used to actually determine that we are going on the right path to save energy.

Guy Daniels, TelecomTV (07:48):
Great. Thank you very much Joan. That's a comprehensive list of standards and organizations there for everyone to be aware of and interesting how there's a lot of linkage between them as well. And what else sort of comes off the back of it Sarat? So let's come across to you as well for your thoughts on this one, about what standards are emerging to understand the energy consumption, especially of network functions and service level work.

Sarat Puthenpura, ONF & Linux Foundation (08:16):
Yeah, thank you. So I can talk about this in the context of open brand where we are being doing substantial work to create what are the best practices of measuring energy consumption on a component base and also end to end. And I must acknowledge that NDIA under the US Department of Commerce as awarded a such grant last year to ONF and we have been doing quite a bit of work in collaboration with Rutgers University WIN Lab in New Brunswick, New Jersey. So the goal is to essentially touch upon the points that Joan mentioned in the context of open ran, what is the best way to measure energy consumption on a component base? Hardware? The hardware, main hardware in or physical network function is RU the radio and then the supporting software functions virtual functions like CUDU and all the way to the core. So the work that we have been doing is to get what is the best way to measure with the minimum effort, what's the minimum data set that we need to collect from the network using or unspecified interfaces, which is not enough.

(09:38):
As Joan mentioned, we need to depend upon other industry standards like Kepler for ban and sc, Andre Redfish and bunch of very similar industry standards. Combine that with or interface and create a good mechanism measurement strategy and approach and using the data can you create appropriate metrics and KPIs. So eventually what we want to do that how energy consumption varies as a function of load, the load means it could be the amount of volume and also the mixture of the type of load that is offered to the network. So the ultimate aim is to create a power model, a function which captures the energy consumption for carrying a certain amount of load with given prescribed quality of service and then use that power model to optimize this energy consumption. So basically we observe measure and then using and model it and with the model we can optimize it a little bit more rigorously. And also our aim is to eventually create some open source functions that could be put in the SMO or R AAP level to accomplish this so that it's been available as a valuable resources to assess the energy consumptions and modeling. And we hope that we are plugged in with the online community to standardize these approaches and eventually the industry adopts us and acceptable standards. So that's my comment on additional comments on that.

Guy Daniels, TelecomTV (11:22):
Sarat, that's very interesting. Thanks very much. And it just shows the amount of work that's ongoing and the attention to this area. Francis, would you like to add some comments to this first question?

Francis Haysom, Appledore Research (11:35):
Yeah, so Sarat mentioned, I think the very important thing is the Kepler initiative in terms of which stands to be clear, Kubernetes efficient power level exporter, which is very much looking at this whole aspect of software, how you make better software. I think the thing that we haven't actually considered, which is as important in terms of this standard is going beyond the telemetry. What do we do with that telemetry afterwards At Q Con last year it was a very interesting presentation by Deutsche Barn, which is the German railways systems and one of the things they were looking at very different than Telco, they were looking at what is my actual footprint of energy usage on the public cloud? And they couldn't answer that question to start with and they started using Kepler to do that. But the important thing that what they found was that what it enabled them to start doing was actually enabling and empowering their development community to look at energy efficiency in the way that they're writing the code.

(12:38):
Because at the end of the day the efficiency is often in terms of are you using the right algorithms in your code? Are you containerizing it in the right way? And this information was less about it being KPIs, it was means by which developers could optimize their code, look at where things are changing. The other thing they found was useful in terms of he described it once it runs nobody cares, is that it allows them to make real time decision making. As things are changing, they can change their code and feed things back. The third thing they noted was that you can start to drive out what they described as better safe than sorry, which is when you are constructing a power budget, quite often each person is adding to a little bit of overhead to the power budget in terms of the way they're developing code and therefore you are building in multiple numbers of over budgeting of power.

(13:47):
And again they found again good telemetry information coming from capital was enabling them to slim down that budget and optimize that energy. So I think I really stress that it's as much about not just measuring it, but what do I do with that? And the final thing I would say particularly for telco is that it's all very well measuring this, but unless we can feed this back into our development supply chain and we can consume things on a much more rapid change basis, all of this information doesn't actually change a lot, change a lot. So it's important in terms of how we can orchestrate this onto the system, how we can build DevOps like agility into platforms that can use this information to drive energy efficiency in the network.

Guy Daniels, TelecomTV (14:39):
Thanks very much Francis. And how often have we said look to other industries and what they're doing and apply some best practices from there. Great example. Thank you very much. Another view of question we have received, it's slightly related. We talked about collaboration already. This question asks in which areas will industry collaboration produce the best results to help us get to the green Ran? Francis, let me come straight back to you on that one if you don't mind. What were your thoughts on what sort of collaboration is going to get us there?

Francis Haysom, Appledore Research (15:14):
I think interesting, one of the things I researched for some time now is the whole area of cloud. One of the things I think it's quite often not fully understood about cloud's. Fundamentally what they're doing is massive operational efficiency. And operational efficiency is about how much load can I put on a single CPU, how much can I optimize the power going into a data center versus the power that is consumed within the data center. So this aspect of multitenancy is a very important thing in terms of optimizing energy cost, optimizing operational costs in data centers, in hyperscale public cloud. And I think if we're looking for energy savings, the RAN and in other networks as all it's this area of how do we optimize the maximize the amount of multi-tenancy. We're still in a situation in a lot of networks where we're still, we've introduced cloud, we've introduced virtualization, but quite often we've kind of nailed it right back down to the hardware because we can't trust the true multi-tenancy.

(16:28):
If you look at aspects of just coverage in countries, we are massively duplicating coverage between operators. Now there's energy to be saved by combining that, but that requires regulatory, it has to require both regulatory models and commercial models of sharing resources or optimizing that multi-tenancy. So in terms of the industry collaboration, I think the biggest area that is going to give the biggest impact here is models that both allow us to retain the kind of competition but also incentivize trading commercial relationships, multi-tenancy between operators, sharing of resources, both of the physical layer, but also increasingly at software level, the cloud level, et cetera. Those are the ones that are likely to drive the greatest benefits in terms of energy.

Guy Daniels, TelecomTV (17:27):
Great, thanks very much for that. Francis, have you got any other comments on this question about collaboration? If not, we'll move on to maybe Joan. Yeah, come on Joan, come here for some comments on this one about what sort of collaboration will get us

Joan Triay, DOCOMO Euro-Labs & ETSI (17:43):
There. I want to emphasize also maybe one point, it's just not only about collaboration within the industry, I think it's also quite important that cross industry collaboration, so telecom industry cannot just do itself to become greener. And for that we need also collaboration with the energy suppliers. I would say that here, one of the points to emphasize is not just only about the contractual aspects on where the energy is purchased or what type of energy is being purchased towards trying to use more green energy, but also about the exchange of information between the two. And for example, from some of the use cases that we are evaluating in industry standards like in 3G VP and other open run and other organizations, there is no possibility to actually perform certain actions if we do not know exactly what is the source of energy that is being used for power that run site or this data center. So actually some of the variables that are being put into these AI models to determine best policies to save energy or minimize the carbon footprint required that we know actually where that energy is coming from, of which type. So yeah, I think that is also important to make sure that there is a good cross-industry collaboration because in the end telecom on its sound is not going to solve it.

Guy Daniels, TelecomTV (19:37):
Yeah, absolutely. Joan, thank you very much for that and I'd certainly like to see more collaboration with the energy industry and it's something we will be addressing in June at the DSP leaders world forum. We're going to have a session on energy. Sarat, let some cost to you for comments.

Sarat Puthenpura, ONF & Linux Foundation (19:52):
Yes, I want to add on to what Francis said on the efficiency of applications that's running in the network. That's very important aspect because you see the emerging of new architectures and then more computing is taken away from UE to the edge. So I think it's very important to have to give guidance to the application developers by the network operators, how they could create energy efficient applications. So that's an important aspect that we should not forget as we moving into new network architectures and new paradigm for paradigms for providing service to the end users.

Guy Daniels, TelecomTV (20:38):
Yeah, you're absolutely right. Thank you Sarat for those comments. Okay, let's move on to our next question then, and this next audience question asks the question, what are the most important factors to address climate impacts from the network? Is it phasing out all the equipment moving to green energy, new architectures changing user behavior? Sure the reality will likely be a mix of all of them, but what are the priorities for telcos? So San let's come across to you first. It probably will be a mix, but where should telcos perhaps prioritize?

Sarat Puthenpura, ONF & Linux Foundation (21:19):
I think it is a complete laundry list of things that we could potentially do. So let me talk about, except one thing which I don't know whether we could be very effective is changing use of behavior. I remember in the olden days when you have network utilization to be increased, used to play with the price, that means the peak period, you jack up the price and then at the VS in the morning you give a lot of discounts. So basically that attempt has been to avoid peaks and trough and make, I would say ratio between the peak and trough will be as low as possible. So in other words, you have the capacity closely following the demand. So price used to be a level, I don't think that will work anymore. And also changing customer behavior don't know how it is. So I would flip the problem other way that you follow the demand as closely as based on the customer behavior and try to improve the network utilization and remove the idling periods as much as possible.

(22:31):
So if you believe in that way to proceed, then the multiple aspects we need to look into. One is the architecture and also algorithms, applications that would accomplish what I just mentioned because the capacity following the demand as closely as possible, that's where the AI comes into picture. You model the customer behavior or predict the customer behavior and put enough capacity and just enough capacity in the network serving the particular customer base. And I think those were the new architectures where we could incorporate AI and optimization techniques would be very, very relevant. So my take on this one is that you create new as we evolve the network and the behavior, sorry, the kind of traffic that's the network I is you evolve the architecture to satisfy that and put new techniques to avoid the network idling and then by which we'll translate into operational efficiencies and energy costs. So the trouble here is that typically the mobile traffic is temporal and position varying and time varying and we are trying to meet that one with a fixed network in general base station. Now we have some advantage in taking the network virtual function, supporting the fixed network. That's a little bit more fluid we can put with the network virtualization, we can address those backend capacities. But I think the architecture, the intelligence to understand the customer behavior and applications which drives the energy efficiency is probably the way to go in my O opinion.

Guy Daniels, TelecomTV (24:38):
Great, thank you very much. Sarat. Follow the customer behavior, follow customer demand. And Francis, what are your thoughts? As Sarat said, it's a big shopping list there of needs to do, but where should the priority be?

Francis Haysom, Appledore Research (24:54):
I think there's two I would pick out here. I think in the short to medium term, I think the new architectures is key. It's the new architectures like more use of containerization, ification, they're going to give networks the ability to change the way they deliver the network to scale up and scale down to heal in much more innovative ways than current networks, which are still fundamentally based on boxes. That doesn't mean the boxes at the edges go away, but a lot more of it can be a lot more flexibly deployed on software with multi multitenancy I've mentioned before. So I think that's one key area to look at and just generally making more use of the programmability of not just the cloud but also of the physical devices themselves, which is increasingly coming I think in the medium to longer term, the user behavior, as I mentioned, the user behavior I think is quite important at the moment we still fundamentally use consume mobile based on really the models that we were developed by the old phone companies in terms of making calls and the mobile companies in terms of making calls and data, we don't really have any representation of when things are costing more, when things are costing less.

(26:22):
And if you look at other industries, be it Uber in taxis or be it in airline companies, there's much more idea of surge pricing or demand pricing in terms of the usage of the network and I to continually build the network and make it bigger and bigger and consume more and more energy for more and more videos which we all consume, et cetera. It's another thing to say actually at certain points we're going to limit some of this because it's important in terms of energy efficiency of energy efficiency of the network. So I think user behavior both in terms of helping that user behavior, allowing you users to make intelligent choices about how they're using the network, but also in terms of incentivizing that in terms of maybe monetary benefits or just feeling good about yourself, benefits in terms of the network will be become increasingly you can't just continue to grow and grow the network without some impact in terms of just ever escalating energy usage.

Guy Daniels, TelecomTV (27:40):
Great, thanks very much for those comments Francis and Joan, let's go across to you. What are your thoughts on this question of where telcos should perhaps prioritize?

Joan Triay, DOCOMO Euro-Labs & ETSI (27:51):
I agree also with what Francis said, that the architecture plays a key role and I think that the use of the open run and more disaggregated network is essential. However for me that is already I would say the state of the art. So we already know that we have availability. I mean we have already this architecture upfront standards are building specifications for future implementation in the industry to follow following these aggregated architectures. So for me that is already like the state of the art. So I don't think it is the next step for where we actually the telco industry can get much more benefits. I doubt that in the future the architectures will radically change that will really signify a measure benefit in terms of saving energy or becoming or making it more energy efficient. Now I would say that where we can actually the operators actuate and have an action, an active action on making our network greener is on adopting new technologies.

(29:15):
Not the architecture itself, but actually the technologies that are being used to implement the solutions according to that architecture. That means that we might need to phase out certain equipment, we need to look for more efficient, power efficient and energy efficient CPUs. We need to look for enhancements in the protocols themselves. Like for example, what is being done in the radio access technology. These are actions where we can actually take an active role. And the second one I would say is a little bit related also on what I mentioned in the previous question on industry collaboration, trying to promote actually the use of more green energy. And these are places where I believe that the telcos can actively participate except for delivery of an output.

Guy Daniels, TelecomTV (30:06):
Thank you Joan. Thank you everyone for those comments. And before we take another audience question, we do need to check in on our poll for the Green Network summit. And the question we are asking you this week is how can telcos most effectively reduce their energy consumption levels? And you should be able to see the realtime results here. They come right next to me here, the realtime votes right here and it does look like sourcing more power efficient technology stood out in front just ahead of sunset old networks, not much love for the public cloud today. Well if you have yet to vote, then you are running out of time. We are only keeping the polls open until the end of today and then we'll analyze the results and my colleague Ray La Maistre, editorial director at telecomtv, he'll reveal the final figures during next week's extra shot program. Right time for another question and it's a big picture question. This one, let me read this out to you. Looking at the big picture, will the increase in usage from connectivity and virtual solutions like working from home, will it offset the increases in network energy use? Ultimately does the green network even matter? And a few seconds ago we got another question in along very similar lines about how can networks drive broader environmental benefits such as reducing the need for travel. So a lot of interest in this area. Francis, what is your take on this?

Francis Haysom, Appledore Research (31:47):
I guess I'd start with some of the comments I've made earlier before, even if it's work from home, et cetera, we still need to be able to change our network to adapt to be the most energy efficient it can be. It's one thing to say I'm saving my work from home for example. I'm saving myself a commute every morning. I'm saving myself a massive office space for the company that I work in. I'm saving myself an awful lot of airline travel that I used to do when I worked in the vendor community. That's all I think baked in. But at the same time, you need that work from home environment to be as optimized as you possibly can and for the network to have adapted itself in the best way for that type of working. And interestingly for working from home, a lot of that is actually putting it directly onto the fixed network, the broadband network, which inherently does not have these huge transmitter arrays.

(33:03):
It's a lot more energy efficient, it's a lot more energy efficient in terms of the fiber. So I think a lot of those benefits are already there. But I think the balance is you still need to be developing the most optimum network. You need to be adapting and adapting to user change, behavior change, whether it's in the day, whether it's through the week, whether it's through the year, et cetera, that's still important. You can't get away from not having an energy efficient network, otherwise you'll just create another energy sink. And I think the other important thing is for that energy efficiency with ai actually one of the bigger, we're increasing using AI to drive our work from home environments. All of our work has been driven by that, that has the potential to actually massively increase the energy usage that is actually being driven by us working from home. So networks need to be as efficient in terms of deploying ai, optimizing ai. So I don't think we can take our finger off this problem and just feel good about working from Barron.

Guy Daniels, TelecomTV (34:21):
No, thank you very much Francis. The green network does indeed matter. Sarat, let's go across to you for your thoughts as well.

Sarat Puthenpura, ONF & Linux Foundation (34:27):
Yeah, I wholeheartedly support the argument that Francis had. Basically what is driving the network usage of energy consumption is at least the past 20 years is applications. I mean they used to be only voice before network, telecommunication network is predominant, but then the machine started talking to each other and then around at least on nineties, the data traffic or I would say surpassed the voice traffic that the turn continues. So you have some killer applications. I think those applications will be coming out the way, at least the A is going, there'll be more applications and there'll be bandwidth hungry and then it'll be consuming a lot of energy. So I think we should be completely aware of the fact that these kind of computational and network intense applications would drive the demand and we got to be very careful in preparing our ourself for the future to have energy efficient architectures and strategies of network operations to put a control over this. So you'll get into your runaway situation.

Guy Daniels, TelecomTV (35:42):
Yeah, thank you very much sir. Thanks for that. And I'm going to come back to you straight away actually for our next question first because the next view of question we've got is all about open ran and it asks, does open RAN show any tangible improvement on RAN energy use and can the R deliver savings or optimize performance? I think we'll go through all our guests here, but Sarat, let's start with you.

Sarat Puthenpura, ONF & Linux Foundation (36:08):
Yeah, so we do fair amount of work in this area. So if you look at open ran as a aggregated architecture where you take the base station and split that into a radio unit and then something called distributed unit and the central unit. So it's basically a physical function supported by virtual function. So if you can look at prima phase, this would increase the energy consumption, but it provides a lot of levers to control the energy consumption when it comes to applications, which you need quite a bit of network capacity and I would say network functions. So on the whole, if you just look at the present situation taking an no B or B or GOB and then split into this with other computing infrastructure added to it, you can definitely see the energy of this will be increasing. But with Rick, there are two versions of Rick, which is near realtime brick and the non realtime brick and the application within will be able to control the energy consumption in a surgical fashion when this technology matures. So we should not get alarmed that okay, this architecture immediately is not providing any benefits say to this one, but, and the technology and the application develops with the power powered by AI is rapidly evolving so that it is a good mechanism to control energy consumption in the network as there is plethora of computer intensive and bandwidth intensive application applications and applications emerge and that's going to happen. So we need to continue to invest into perfecting this technology based on this open plan architecture and I strongly believe that's the future.

Guy Daniels, TelecomTV (38:20):
Great. Sarat, thank you very much. Very clear call effort for continuing the work and investment. Francis, we've had about three questions related to this topic and this is probably the most representative question of the lot. So what are thoughts on what Open ran can do and the Rick in particular?

Francis Haysom, Appledore Research (38:38):
I think from a purely theoretical point of view, open RAN and Rick can deliver these savings. It's inherently built to allow innovators to create new ways of delivering the controllers that are controlling the ran to optimize the efficiency. As I said earlier, it allows the best developers to create the best code for the best outcome In terms of energy in the ran, I think practically maybe in the short term we have a little bit of an innovator's dilemma with Telco, a lot of the behavior has been to do a, what I would term is the initial proofs of concept and trials with a lot of new innovators. But recent experience I think certainly in terms of as the CSPs have moved open, ran into the mainstream as part of their major overhaul, much of the evidence currently at the moment is that it's going to the traditional vendors backed up by very compelling price, price performance.

(39:50):
In terms of that, yes, the vendors are innovators, they're providing new capability, but they're not the whole spectrum of that innovators. And I think if Telco wants the best innovators in energy in the R, they're going to have to allow that environment, that ecosystem to develop and dare I say, they need to be investing in that ecosystem, not simply sort of leaving it to hang out to dry when it comes to the big procurements because the small players will wither on the vine if that occurs, it simply becomes a scaling and selling opportunity for the legacy to hold out on that. So from a practical point of view, that's thing if the CSBs want the energy savings, they want the best innovators in the thing, they need to create an ecosystem and a procurement environment in which that innovation can occur.

Guy Daniels, TelecomTV (40:56):
Thank you very much. Francis and Joan, Open ran all about innovation in a way, opening up the interfaces, allowing innovation at different levels there, different areas. The development of the R, is it showing any tangible improvements for energy efficiency and optimization?

Joan Triay, DOCOMO Euro-Labs & ETSI (41:17):
Well, it's a little bit early to say from the actual field trials. So many operators are still in the process of deploying the open run and then to be actually capable to comparatively assess the energy savings. But if I go a little bit also like Alad mentioned, from a theoretical point of view, definitely we should see benefits and as we know the run is the part that concerns the most in the network. We also know that the radio units are the ones within the open run that also contribute most to the energy consumption by the fact that we are with this open run, being able to disaggregate the network de couple the radio units from the rest of the network, we can address the problems that we have there in a more particular manner without open run that would not be so straightforward. I would say that also comes out from the benefit that such this aggregation enabled to decouple the problem and use the best of the breed technologies that you can get in each of the different areas.

(42:38):
Whether it is on the power amplifiers, whether it is specific to the radio part and whether it is to the part that concerns to the virtualization and cloudification for the rest of the network functions. We also have lots of other new technologies there and mechanisms that we can save energy with. Consolidation of workloads in the data centers, changing the power states of the CPUs and so on. So without knock and run, I'm not so sure how we would be able to tackle these problems on an individual basis taking into account of course that's just the problem is there's not only one. I mean we need to address and solve the problems all of them, but without being able to identify them as following the open run system, as I said, I'm not so sure how it would be possible. So I'm betting that is going to be a tangible improvement in the future.

Guy Daniels, TelecomTV (43:42):
Okay. Positive news. Thanks very much. That's very interesting Joan. Thank you. And let's go back across to Sarat for some additional comments.

Sarat Puthenpura, ONF & Linux Foundation (43:49):
Yes, we have some studies along the lines with Joan mentioning about how much of energy can be saved and what are the major considerations. So in under Linux foundation leader projects, we do have a program called SMART 5G, and we have been creating an ecosystem of vendors and operators to study this one. So when it comes to energy saving, there are two competing factors. Now much energy you can save from the land, about 70, more than 70% of the network energy consumption. And what is the quality of service? So if you want to put quality of service the highest and the energy secondary, there's one discussions. And the other one is if we can at certain loads we can compromise a little bit on the quality of service and based on the service level offering to areas customers, then we can have more deeper to energy savings.

(44:52):
So the experiments that we did without touching the compromising or touching, compromising the quality of service, we can easily cut 20% if you do it carefully using WIC based application sex apps and predominantly our apps which is sitting in non realtime R. And also one advantage is architecture is that if you look at a desegregated architecture, the R and the application has visibility on a cluster of radios. So as Joan mentioned, significant amount of energy is consumed by the Radiohead radio units. So you should be able to do a holistic, I would say optimization or balance between quality of service and energy saving by shifting the load around from one fuse in a cluster of a base station as opposed to an individual one. So with all this combined, I think the desegregated architecture does give more opportunity to do energy savings and should be able to carefully balance between quality of service and the amount of energy that can be saved in the network.

Guy Daniels, TelecomTV (46:05):
Great, thank you very much Sarat. I think we are approaching the end of the program, but let's just try and squeeze in one last question from our audience and maybe some quick answers for this one. And it's a question about spectrum. What is the role of efficient spectrum assignment for energy efficiency and what could the role of A IB in this context? And Francis, I wonder if I could come across to you if you've got any thoughts on this.

Francis Haysom, Appledore Research (46:34):
I think that as basically if you can combine spectrum, you can optimize spectrum bands, you can save energy, you can be more efficient in terms of your coverage and the way you deliver the ran. I think the interesting thing is how do we enable that in an environment in which we we're very fixed? A CS, B bids for spectrum, it owns a set of spectrum and we're not necessarily used, we're very fixed in that environment. Several conversations with a company called Federated Wireless is looking at the idea of how do we more efficiently trade and combine spectrum between providers and in terms of the new spectrum like CBRS, I think if anything the energy comes down to commercial model. How can we get better energy consumptions by collaborating as an industry to get better spectral efficiency, better arrangements. A lot of it comes down to changing or enabling different commercial models between owners of spectrum, bands of spectrum, geographically in time, et cetera, et cetera. So yes, there are opportunities here, but I think it largely comes down to rethinking the model. Again, looking at the commercial models, how do we both allow CSB to maximize their spectral ownership but also to trade and manage that.

Guy Daniels, TelecomTV (48:20):
Great. Thanks very much Francis and Sarat, we'll come across to you as well for your views on this question about the role of efficient spectrum assignment.

Sarat Puthenpura, ONF & Linux Foundation (48:29):
I have a quick addition to what Francis mentioned. So spectrum has got an important bearing on the energy consumption. So for example, if we have a spectrum band which has got larger coverage as opposed to millimeter waves or smaller footprint, you need more base stations and typically the energy consumption can go up. However, if you have multiple base stations and if you can have a mechanism by which you can turn on and off these pockets will again, capacity following the demand there is opportunity to control the energy usage at the same time satisfy the demand and quality of service in an efficient manner. And also if you look at the transition, I mean from LTE or 4G to 5G LT is bandwidth of typically about maxa for 20 megahertz. Whereas if you look at millimeter waste and 5G can go to 400 megahertz and they do a mechanism called bandwidth parts to reduce the energy consumption and the handset. But that puts a toll on the base stations and we need to carefully manage that. So what I'm saying that the spectrum has got an important connotation on energy consumption and we need to be aware of putting in mechanism to manage energy consumptions under these kind of scenarios.

Guy Daniels, TelecomTV (49:56):
Fantastic. Thank you very much for your comments there. We are out of time now, unfortunately. So I'd like to thank you all very much for joining us for this live program and that's a wrap for this year's Green Network Summit. Thank you to all of you who submitted questions. We did try and cover as many as we could in the time we had available. You can watch all of the programs from this year's summit on demand from our website featuring this incredible group of industry experts. Thank you also to our speakers and sponsors and viewers for supporting the DSP Leaders Summit series. And we will be back next week for our special Extra Shot program when I'll be joined by telecom, TV's Ray Le Maistre. And we'll take one final look at the highlights from the summit, the main talking points, the key takeaways, and of course the final poll analysis. And for those viewers watching us live, we are going to broadcast today's panel discussion immediately after this program. So do stay with us. The next topic for our summit series is Telco as a platform, and you can join us for that in April. But before then, of course it's MWC Barcelona and it's the return of the slice, our comprehensive daily coverage of the biggest event of the year. For now though, thank you for watching and goodbye.