At the cutting edge: AI-driven sustainable 5G networks

Evening. Thanks for taking the time.

You know, customers across the globe, if we think about the telecommunications industry, have been looking at creating differentiated solutions - AI driven networks. There is a big urge to drive differentiated offerings that they can provide both to the consumer and businesses and taking a digital first approach. That's what this talk is about today.

Uh welcome to TLC 302. I'm Robin Harwani. I lead the global technology teams focused on our worldwide customers and partners in the telecom IBU at AWS and with me, I have Lana.

Lana: Hi, I'm Lana Glock in Iran, North America telecommunications solutions architecture for AWS. Thank you guys for coming. Hello. Awesome.

Robin: La. Uh just to kind of set the scene. Well, Lana, why don't you tell the team what our customers are telling us globally?

Lana: Sure, I'm afraid right now our customers are telling us we are standing between them and the happy hour. That is not the position I like to be in, but we'll make the most of it.

Um so as you can imagine, we work with many global communication service providers or CSPs to help them transition um their infrastructure to digital and AI driven transformation. As we work with them, we see quite a few patterns. Even though the telcos we work with are very different, they're different sizes, they are the different cloud maturity levels and so forth. There's still certain things that are common across all of them.

Um first telecommunication companies are creating um and accelerating revenues through new revenue streams that are enabled by the new technologies such as 5G um and AI/ML.

In this industry, telecommunication, like many other customers that we have tried to improve their profitability and for that, they're looking to share their capital expenses and their operational expenses and that they are doing by reducing their hardware on premises, footprint and migrating their uh their um applications and network functions to the cloud by leveraging generative AI technologies such as chat bots and recommendation engines.

Uh telcos can provide hyper personalized services to their customers as well. Um and they are delivering customized products and services to them.

And finally, we see telecoms becoming strategic partners for many different enterprises that are looking to innovate. Uh telcos are leveraging infrastructure connectivity and their expertise to provide 5G and IoT technologies to um companies across the globe.

And from this, it is pretty evident that choosing the right partners to do your transformation with and getting the right enablement is key to the business success of our customers.

Thank you, Lara. I think it is critical as you can all imagine across these goals. One of the questions we get asked by our customers, how does a CSP plan the journey to be a digital telco and have a cloud native, sustainable AI driven 5G network?

That's a lot to achieve in 60 minutes, there is a lot of deep dive architectures we are going to get into. So we look forward to that uh diving in with you guys, but to level set, what we are going to cover today is five domains within the telecommunications networks.

The first, how do our customers, silicon, customers create differentiated solutions both for consumer business as well as the B2B2X, the business to business going into consumer or another business solutions. What kind of frameworks do they use? How do they operate across different ISVs partners and the other operators that they need to interoperate with from a roaming perspective, from interoperability perspective?

Then we'll dive into customer experience. Lana will walk us through that going into the customer interaction layer, created differentiated customer experiences and making sure while they do that, taking a cloud first cloud native approach.

Then how do they modernize their OSS and BSS systems to make sure that these systems are ready to offer, bespoke as well as repeatable offerings for consumer and businesses?

Then we'll dive in to the core, the the network core and the access network. So there is a lot for us to cover and to kind of set the stage at a very high level 10,000 ft, if you think about it, setting up an entire network, if you were here a couple of years ago, this was our framework.

We set the entire, we have the regional infrastructure from AWS that can be used for deploying applications across various availability zones in an in a highly available resilient performance, uh optimized and secure environment.

Then we have our Local Zones and Outpost infrastructure that allow for workloads that require high data rates as well as low latency to be placed at the edge of the networks. At the same time, use cases that require applications to be deployed at the edge can be can be deployed within these Local Zones and Outpost at the edge of these 5G networks.

And finally, at the far edge of the network, you have the virtualized access network components. This is deploying at thousands and thousands of radio locations, deploying virtualized infrastructure. The same way our customers have been deploying infrastructure in the cloud and at scale deploying applications, the same observable framework they can leverage at the forage.

But since we launched this framework two years ago, we have learned a lot and there are many vectors to think about uh Lana. Do you wanna take the team through that?

Lana: Absolutely. Thank you for covering the 100 level uh view of this. However, we cannot use the same framework for all of our customers across all the operators. They must factor in multiple dimensions.

Uh then uh we help our customers define their journey um to the CSPs across the world.

Um so some of the things to consider as you are embarking on your journey is, first of all, is your CSB a connectivity focused primarily providing the underlying network infrastructure and basic connectivity or are you a full step CSB providing a number of other services and offerings and the whole suite of potentially addition, additional products and services on top of your network.

In such case, if you're the latter, then you're going to have to think about your customer experience a lot.

Um are you a multi multi country telecom like Vodafone or Orange? For example, operating across different countries. In such case, you would have to take into account the local regulations and potentially local data sovereignty rules uh which is a pretty big consideration as well, especially in Europe.

Um not every country is going to have an AWS region either it doesn't mean that country cannot deploy an AWS. It just means they will have to use a different region. In addition to that. Uh not every region is going to have the same suite of services offered uh in AWS. So all these are things to consider for global companies selecting the right partner is another critical uh thing to think about critical decision to make AWS has hundreds, if not thousands of partners.

Uh but you will need to select the partner that makes sense for your business. Maybe you have an existing relationship with an ISV that you want to build on. Um and that would probably play into your uh partner selection as well.

In addition, telecoms have different cloud maturity levels. Some of our customers are, are fully enabled in the cloud and understand as much as we do probably about certain aspects of AWS while others are just beginning on their journey and where your company is, is going to play a pivotal role of how you're going to start your transition.

In the end of the day. It's going to have to make sense for your business. So where does it make sense to start? What can you do in parallel and what can bring the most value to your business? So these are all the decisions that you're going to have to think about while making your transformation.

And here today we are going to try and help you frame um the way to make these decisions.

Thank you, Lana. So we covered the 10,000 ft. We we we dive in, understand the vectors like multi country operators, different journeys within the ISV ecosystems, the organizational readiness. And then once we dive in to the first one, how do, how does an operator start and build an entire solution that helps them differentiate across their consumer space across their SMB domain as well as the enterprise domain.

And why does the operator i have the right to play or unique position to, to capture this opportunity. If you think about it on this chart, at the bottom of this chart in each consumer or a commercial uh customer, there is a connectivity provider on top of that connectivity layer which is powered with with many many different types of devices across uh different protocols.

Makes models you have the entire ecosystem of, of devices and services over the top services that are powering up value to your consumer as well as the the uh enterprise customers. The common value that that each home has or each customer has is the CSP the connectivity layer that has visibility from that home router or from the connectivity of the home into the entire set of devices, services that are uh geared up.

And that we believe is a very unique opportunity we've been partnering with uh customers like Tellus that have helped us to create how we can create uh formulate a framework that can be used across the industry on AWS to create offerings like this.

So let's dive into the framework. If you think about this, this framework, we have worked with TM Forum, uh the ODA the Open Digital Architecture framework team as well as very closely with our customers and our partners to say, how do we create offerings that provide B2B as well as the B2B2X solutions while reducing the complexity across a fragmented IoT ecosystem, fragmented connectivity device and services ecosystem.

And so at the bottom, if you can see the connectivity in the network layer under which there is dozens and dozens of devices and we will dive deeper into it. as we, as we go on top of that, there's AWS services that allow for our customers operators to, to stitch together various unique CSP services which is then exposed with a common API exposure layer, very, very similar to what you would think of in the ODA um framework.

On top of it, customers are on the common platform. The CSPs can provide consumer offerings, SMB offerings and the enterprise user offerings. This is where there is a completely differentiated transformational approach that we believe instead of having separate monolithic systems that offer bespoke value to the customers in a horizontal approach and stitching together in alliance with the the standards bodies uh and taking the best practices and, and giving this framework to the industry top of AWS will, will enable a new set of ecosystem innovation.

Now, if we take an example of taking the the framework and apply that to our home, this is the work that we have been doing with the Tellus team at the very bottom, just like I said, devices layered, there is cameras, thermostats, there is locks, um like smart bulbs, lights controllers, so many different device devices that integrate into AWS through the AWS IoT portfolio.

Now you will observe here that the connectivity layer is abstracted out but you can you can very well assume it's either a fixed or mobility network that is powering up the connectivity here. From that point, there is different services that that can be used such as API Gateway, Lambda, Clean Rooms for data management, the Telco Lakes uh services that we can pull together from a Lake Formation perspective to create services that are then offered by the operator to uh achieve value.

So what could be those services? Things like providing home security, convenience, skews things for energy management along with the unique things that operators network enables them, you know the identity, the omni channel experience, the care few services that they can offer experience optimization, notification management, all of these capabilities with a single bill and care framework that can be offered to, to customers.

And that's what we believe is the digital life uh solution that that can can power up not just the consumer experiences but also the business experiences.

Now there is multiple layers i covered, you can see the first and the second layer. uh if you map it to the framework we saw on the earliest slide are are the devices the 1st and 3rd party devices, then you have the AWS cloud, the CSP middleware and then the smart apps that you have.

But so far we we haven't covered what are the what new launches we have done? uh how does that change? uh the the platform now here, generative AI enables to, to offer additional value, completely transformative experiences and different um different new uh offerings to, to in this space.

For example, if you are going into the home, looking at device data across your health devices, whether it is Fitbit or or Apple devices, different devices you can as an operator provide value to uh consumers for elderly care, anxiety, real time, translation, anxiety detection and and taking actions.

There is a lot of routines that can be offered and we proactively uh recommended to the consumer based on the the generative a i capabilities that that we can think about. So how does the reference architecture look like if we go even deeper?

So from uh from right to left on your chart, you can see you have third party clouds, different cloud providers that enable for for these devices in your home to operate, then you have your consumer premises. Where do you have bunch of gateways iot hubs, echo devices, hopefully a lot of different user devices that are there as well working in conjunction with the existing telecom network, uh which is a fixed network or uh a fixed and mobile network these days, a lot of fw a offerings as well.

So that integrates into the aws cloud then integrates into the zero type solution that we have. Uh we have built along with the experience controllers integrating into the aws iot uh framework, uh all of the data both across the telecom data as well as health data can be uh persisted and managed across uh the the lake formation uh uh services.

And then you have uh the application middle layer. The application middleware is what the operator either licenses from a partner or builds itself to create differentiated offerings like identity notification orchestration, like we discussed about uh in the previous chart. And then that is augmented by the generative a i capability.

So if you think about it generative a i um with the, the previous framework enables these services to be then offered to the common marketplace that the telecom uh consumers are looking to uh to, to avail with a single bill and a care framework.

If i, if i take this to the next chart, these services then can be offered both to consumers uh as well as enterprises. Now, the last bit that we talked about in the, in the framework was the marketplace a frictionless experience for customers to bundle together offerings from multiple operators, multiple isv partners as well as uh and end users.

Uh and that's the the marketplace solution that can integrate into aws marketplace through service catalog as well. We are doing this with multiple operators, onetel being one of them. Thus, uh n tt a one, just a few examples where we have, we have this platform life.

So why do we do the same thing if the customer is in enterprise, the same marketplace, the same framework, how do we extend that all the way from offer orders, usage and, and service? So here this is another example where we are doing this with uh an operator uh in, in north america.

We are doing, we are working with sales force nokia for charging our own wavelength service team as well as amazon.com integration for retail ser uh retail hardware bundles that can be packaged into the offer that goes out to the customer, all of this to create a new five g mac uh experience uh with, with our isp partner summit to give a uh seamless uh vr experience to, to the customers.

So that hopefully provides you a little bit of uh context of how an operator can create, leverage a framework, create offerings, a platform which can be used to buy different lines of business to offer services uh uh and, and differentiate including how generative a i plays a role there.

So lana, do you want to take the the team through the next bit in terms of customer experience? Absolutely, customer experience is one of the top factors um impacting customer loyalty. This is what 95% of our customers tell us.

Um this covid-19 pandemic, our expectations as customers of seamless digital customer experience have definitely heightened and um we are very quick to move on when we don't get the digital experience that we want uh right away to keep pace with customer demands. Many companies are leveraging new technologies such as five giot, gen a i and a iml uh to deliver excellent customer experience.

Um these enterprises are relying on telecommunication providers for the network speed and for the infrastructure that is required uh as well as the bandwidth that is required to support these new use cases. However, when we think about telecom, this probably is not the first thing we think about when we think stellar customer experience, uh telecoms in many cases struggle to provide that to their subscribers.

So why is it a challenge? So i will ask you a little trivia question, who knows when the telephone was invented a year? You can yell out any number who thinks that it was before 1900? Good. It was before 1900. That was actually 1876 when the first telephone was invented and believe it or not, the first telco was created uh established even before then. And this telco still exists, exists today.

So bt was one of the first ones. Uh what i'm trying to say is that telecommunication companies are very old legacy companies. A lot of them are large. They've been around for decades, maybe even over 100 years for some of them. And they have a lot they, they have accumulated a lot of legacy technology over time. And you can imagine that and you probably know from experience that innovating in such an environment is not easy written and replace methods don't really work very well if you think about the complexity of the systems that telco has to support.

Um so, cs vs need a more effective ways to leverage customer journey data that they have to guide their customer experience decisions and transformations without having to undertake and complete large scale technology overhauls first.

So what i would like to share with you today is a solution that i was lucky enough to be a part of this solution can help telecom operators improve their customer experience without ripping and replacing their legacy systems.

First, the solution is called connected customer journey or c cj. And it utilizes a federated data lake approach on aws to bridge organizational and data silos in telecoms, it connects to telecom's existing uh data uh storages and systems uh providing real time visibility into uh customer journeys across channels of interactions.

Additionally, the data is also aggregated for further analysis and um making further business c cj platform applies across the customer life cycle, providing a comprehensive view into the customer journeys over time.

So the customer journey is not something that happens, you know, within a few minutes, it could be something that takes days or even weeks. For example, if you try to buy a new phone, uh it's an expensive purchase, it may take us longer to decide. And through that time, we are going to do certain things that we can all accumulate and, and connect.

Anyway, the key benefit of this solution is enhancing customer experience without having to replace the legacy systems that we have in place. And i want to look a little bit at at the um systems and services that we are using for that solution for that platform.

So we are going to start with the data source uh layer where all of the telecommunication data sources uh exist today and this could be on premises, it could be uh on the device um or it says vss systems and maybe call center logs and so forth. All these stay in place and don't need to be replaced.

The second layer is the data ingestion layer and this is where the data from these data sources is going to get uh consumed um and ingested into aws cloud. So for that, we can use several different uh methods and some of them are listed here. Uh we can use kinesis and c fa we can use veg uh loading with glue and aws transfer family if this is what suits our uh our use case or we can use uh api s through api gateway.

Um in this case, all these systems can actually call api s with the events that we are interested in that user is going through and ingest them into um into c cj. So once the data is in the c cj, what do we do with it? We build something that is called connected journey graph using amazon neptune database. It is a scalable um database with high availability. It is also serverless and it is a graph db. What it does is to accumulate all of the steps in a customer journey and connect all of them into, into one journey.

Some of the data, the raw data be stored in uh amazon s3 buckets as well for analytics. And we use like formation for easy access control. We also leverage aws glue to move the data into the graph uh database and into our next layer. And our next layer is going to be the connected customer journey insights.

Um so the system has 22 layers. The first layer we talked about was core and this layer is the insights and this is uh where all the magic happens. Uh this is where data is analyzed and um we generate valuable customer insights that can be used later. And these insights can power uh real time um um real time data that's sent for customer agents, for example, that are talking to the customers. It could power uh business uh dashboards where we can, it can help us make better business decisions. It can power hyper personalized chat bots uh to have interactions with the customer that are within the context of their transaction.

So with that i wanted to dive in a little bit deeper into the architecture for c cj core. All right, let it build out. So c cj core is responsible as i mentioned before for ingesting data from your existing systems into aws. Uh in this case, we are using a stream like kis for example, to ingest that data. So your data sources essentially publish events onto the team in a standard format. These events are published and as they are published, the metadata is also extracted real time by the event processor. You can see it here in orange. Additional metadata can also be extracted by the data metadata extractor once the events are in the database. So you can see that on the bottom at number four. This metadata is stored in the neptune knowledge graph and the knowledge graph trains customer journey models using sage maker. These models can predict things like purchase propensity of a customer or churn propensity of a customer. So basically this is where we gain the insights and then the telco insides.

So what happens here is first um we extract customer segment data and the segment could be prepaid customers or postpaid customers or maybe customers in some uh geographic area. Um we get this data and um we save it into the s3 bucket. After that, we leverage aws glue jobs um getting the data out of the s3 and putting it into amazon red shift warehouse. Additionally, we will leverage customer journey analytics api s to retrieve metadata and insights uh regarding customer journey events. And finally, we build the custom applications, visual uh applications and visualization. using this data inside. We can either connect directly to the red shift to build these applications or we can funnel the data into the external systems like crm systems for example. And this will provide customized reporting um and analytics on the customer.

I am not going to dive into this architecture just because i think i'm jamming you on time and this. But um basically what this shows is how we can use c cj in this case to send a hyper personalized offering to a customer who was the latest event uh connected customer journey decided that the churn potential of this customer just went up.

Um so if you want to know more about the architecture and see the actual demo, you should definitely stop by our demo booth in the venetian. It's booth 580 aws four industries is the uh area where we have a booth there.

So very quickly here basically on the right, you can see the dashboard, it's just an example of a dashboard that we can generate using these analytics. Um and um you know, here you, you can see the out of the user, the c set by product type, average number of channels, switches per day by intent

These are just examples. Um a lot more is possible with this system. But let me ask you this, how are these customer journeys enabled? What are the backbone systems that are running uh behind the scenes and enabling customers go through their journeys? Anybody wants to guess because these are the systems we are going to talk about next. You can yell out S SBS S yes, but you knew that.

Ok, so we'll talk about OS SBS S systems. I want to give a very quick refresher um in case uh you don't know. Um so, OS S systems um help uh manage um the complex technical network infrastructure and hardware and switches and software that services all of that in, in data centers. And some of the key SS functions include network inventory, uh management, fault, and performance monitoring, uh traffic engineering and so forth. So when you think OS S think network and network services, uh network servicing software, now BS S enables the business processes that support customer acquisition um and service delivery. Um so, and, and care BS S functions like product management, order, orchestration, customer billing and invoicing and crm help drive the optimal customer experience.

Um so when you think about BS S systems, you can think about systems that help us interface with our customer. Um today, there is one major development that is having a huge impact for telcos and um it makes telco's reliance on noise s and bs s systems higher i'm going to ask one more question. What is this development? Don't say, what do you think this development is? I'll give you a hint. It's on the slide in big font with a g with a g.

Yes. Right. Thank you. Uh so five g networks are about 100 times faster and about 1000 times uh greater bandwidth than four g. Uh providing telco's tremendous opportunities with new business revenue streams. Uh with five g telcos have a power to capture more customers and to drive growth.

Um so why is there the increased reliance on os s and bs s systems? Uh with the arrival of five g? Well, for 15 g is using uh ra a new radio frequency. It's using dense small cell architectures and um vir virtualized technologies which make os s support for them a lot more complicated and it requires upgrading the os s systems. Uh in addition, with the higher speeds, there is a lot more data that is now going through the network. So for that, we also need uh a better os s and better bs s uh systems.

Um it also opens up a whole array of new value driven businesses. Uh the network slicing, for example, now enables us to charge our customers based on usage, not based on sub sub monthly subscription, for example. So all of that requires us to operate our os s and bs s systems. And this is a pretty urgent need for telecommunication companies that are looking to innovate.

Um if your telecom organization is running legacy os s and bs s, you will not be able to take full advantage of five g. Um so it is critical to embrace new os sbs s models as well as migrate to the cloud.

So how will modernization of os sbs s systems and five g and network slicing capabilities together with aws services and a s and ns delivery models enable telecoms to better modernize their five g architecture?

Well, by choosing the right partners and we talked a little bit about this earlier and we have some industry leaders like ericsson, nokia sales force and docs aws is enabling telecom operators to adopt ce sbs s model. And together we are creating bundled offerings based on um based on usage rather than um standard um offerings that are the same for uh all kinds of uh consumers.

So if you think about a large company who is a consumer versus a small company, who is a consumer, they're going to have different uh slas, they're going to have different bandwidths that they require, they are going to have different uh data, uh amounts of data that is um needed to power them. So that's why we need to create these bundles that are now um centered around um the usage, the amount of usage in slas.

So aws provides the agile api driven architecture required to support dynamic slicing and on demand slicing provisioning. Just to, just to add there, great point. Like since we started r os sbs s partner ecosystem and, and our community of partners that we have been working with since 2018, we've, we've worked closely with, with our standards and, and framework bodies like tm forum to make sure that our best practices from aws are, are then transposed into the, the leading edge catalyst, leading edge uh isp partners and vice versa. We have adopted a lot of learnings from the industry bodies to make sure our isp partners are truly building cloud native uh uh os s and bs r systems.

Now, as we go into the architecture, you will hopefully see how uh the entire frameworks that we have translate into a tangible architecture that you can drive to production uh across all the the layers go for it. I just wanted to add that.

No, thank you. I i appreciate it.

Um so if you look at the architecture, we are going to start at the data center. So on the left, you would see um technology that you would have uh on premises, you may have your active directory or your dns uh your databases and servers, you would connect to aws, whether uh you have a direct connect, uh you can use that otherwise you can use client vpn.

And the next layer that you see here are all the aws man managed services uh that you may use, this is going to be a landing zone and you will see two availability zones uh for for high availability of your applications. You will also see the vpc uh and the uh connecting layer into that.

Um here is where we are going to build your applications uh on the bottom, you're going to see the data layer and uh it's going to use um cross a z replication again for high availability of your applications you compute is going to be in this layer. So whether you're using uh containers or not, you would want to use auto scaling here. And um all your compute nodes are going to be here, your file storage and your ebs storage is going to be in this layer as well.

And then you have your basically your public subnet with the integration layer where you're exposing your api s or the other way around. If you're consuming api s from the from external uh api sources going to be here as well. So this is a pretty typical reference architecture for s sbs s.

And there are a few things that um you would want to think about uh one like with any aws uh deployment or any public cloud deployment for that matter. You want to make sure that you put your um and you think about the cost and cost optimization uh optimizing your architectures before you do anything because uh we don't want to overpay so this is where we want to have our uh autos scale in groups. We want to use serves as much as possible and so forth.

Um one other thing to consider is that a lot of o bs s applications are highly latency sensitive, yeah, highly la latency sensitive and it is very important for some of them uh to be close to our customer. So this is where aws can augment your own premises with the uh cloud continuum. Uh whether it's going to be our regions and availability zones or maybe the edge uh which will bring your data and applications closer to the customer.

So with this ubiquitous cloud continuum enabled by five gh computing bs s functions can be deployed in the flexible distributed uh architecture to optimize performance and latency and costs.

I'm going to go through this quickly because um i think we are upon time. Uh but basically, if you look at this slide on the left, you're going to see all the different telecommunication companies that we work with. And here you will see some of them that are tier one telcos in the us, like t mobile, for example, or vodafone in europe or you're going to see liberty global.

Um this one. Yeah. Uh liberty. Latin america actually, excuse me, uh that operates across five different countries in latin america. There is globe uh and there is a lts, some of these operators um don't even have um don't even have an um a region in the country they operate and yet they're able to deliver uh services to their customers.

Um using aws globe, for example, is uh servicing philippines and is uh utilizing um the singapore region uh to deliver their first a pg network assurance platform deployed in aws. So basically, there is a huge diversity in our customers.

If you look at the uh second column where you'll see um the um partners that they're working with, you'll see some like m docs and erickson salesforce and uh for some, it makes sense to go with others. So anyway, for, in a w us, we have a lot of different customer, uh a lot of different partners where um there's quite a few choices based on what your business uh needs are.

I will go quickly through liberty, latin america use case um liberty, latin america is the leading mobile broadband provider in the caribbean and lam. And they have become this leading global provider by going through a number of acquisitions. They had at least four major acquisitions including uh claro chiles, claro panama.

Um they acquired at&t operations in puerto rico and they acquired telefonica in uh in uh costa rica. So if you think about that for a second, um and think about the variety of voices and bs s systems that they needed to integrate into one. They um it was a pretty huge undertaking.

Uh they decided to go with the uh architecture that leverages outpost local zones and us uh one region. They, if you look at the circle um here, you will see all the partners that they are partnering with for different solutions.

Uh they decided to go with cloud first platform with open a is on aws uh which is a departure from traditional bs s delivery models. Migration to cloud hl methodologies fundamentally transforms the way it can work with business and come up with the new business products for their customers.

Um so eventually when they are done, the bs s system uh in aws is going to support all of these acquisitions. As one another example um is t-mobile uh that was able to implement blue green deployment for their vast um footprint in the cloud that includes over 5000 compute nodes.

Um i'm going to move on here just because i think we have a lot more to cover, but basically to sum it all up. There is a customer journey. And if you think about a customer buying a phone, what do we do? Well, we do research, we then engage with our provider. Uh we purchase, we activate the phone and so forth

Each one of these will have a flip side of operator journey. So while we research operator needs to plan and forecast how many phones, for example, they're going to sell uh while we um get our phone and turn it on, operator make needs to make sure that they provision and activate it.

Um so it's kind of like the um two sides of the same coin. And here in orange, um are all the use cases that we have implemented using generative AI technologies. So we have upgraded the way that um store assistance work, for example, or the revenue assurance is one of the other prototypes that we've done with orange that um has uh worked really well for them.

So just uh now I would like to ask Robin to take us to the core element that is underpinning all of that. So hopefully, as we started, we started at 10,000 ft, we talked about how do we place different workloads? We had that 10,000 ft slide where you know, you could just take a network, you take an entire telecom operators workloads and place them in different places on AWS infrastructure based on data networks and latency profiles.

But as you saw, we walked through consumer and business offerings that can be built with marketplaces that are TM4 mod A compliant on top of that, then we went into the customer journey, then we went into the USSBS framework that we, we have to cover and now we will cover the networks, right? As, as CSPs uh build out their networks, they have few requirements.

This is something that has been going on some time for some time. They want to make sure that building a secure virtualization environment which is private managed by them on AWS. They want to make sure there is cloud application at the edge for applications to be deployed at the edge for data support as well as low latency requirements, ability to segment uh and deploy updates, upgrades and uh and and new capacity of the network is critical for operators.

And while doing that on the same infrastructure delivery CICD pipeline, it is critical our customers like this want us to want to deploy software updates on their radios across thousands and thousands of locations. And then they want uh to monetize this network both at the the central points of presences as well as the at the at the edge by deploying applications.

Now they want the the largest community of partners that can build and deploy these applications at the edge as well. So our approach with them is to extend our ubiquitous cloud to the edge for the telecom workloads, extending cloud to telco edge locations within operator premises like what we have with uh outpost services and for joint good to market, we have wavelength that enables them to do the same thing for enterprise applications to be deployed as well.

Now, if we think about how do we jump into an architecture, how do I build out a 5G core network quickly leveraging AWS. Now here is the um the the architecture. If you look at the, the far right of your slide, you have a region where you deploy your control plane, leverage the management tools that you have available. The same cloudwatch, the same cloudtrail, the system of simple systems manager to update, update and manage the network.

You're using EKS clusters to deploy the control plane. You're extending that to the edge here in this scenario with a local zone and in that local zone, you are deploying the user plane in a COPS architecture and there uh that is then integrated into a telco data center, hopefully at a site where you can use both AWS outpost or one U two U outpost for virtualizing your radio location.

And uh CUD for those of you who are family with that, I'll walk through the access network in a bit as well. The important thing to understand here is the breakout. The local breakout is happening in the LG uh in the local, in the local zone. And you are going through the IGW or an LGWD area to, to traverse the traffic to the internet.

For to deploy this uh this same core platform. You can do that in a primary and secondary scenario, an active active scenario, capacity expansion scenario or a disaster recovery scenario. You can do the same thing in a second configuration here using outposts. And here the difference is the outpost, a full rank outpost can be deployed in the custom premises, customers premises to deploy the user plane of the core network.

See the, the value here is the same observable that you have across the deployment. The same CI CD, we have thought through things like DPDK Malta that are required for uh network segregation and the VPC uh which is extended both in the, in the region as well as to the, to the uh to the edge uh which are, which is uh available now, a live customer.

For those of you who are aware, this has been our customer for the last few years, 15 plus local zones, 16 availability zones across us. Like we spoke about thousands of site specific hardware connected to AWS cloud. We have the same command and control fabric.

If we go deeper into this architecture, like i spoke earlier, there are national data centers where the OSSBS the subscriber management, the front end systems, the IMS core control plane is deployed. Then there's the regional data centers, the local zones where uh where the, the user plane uh uh is, is deployed. And then there is the edge data centers where the UPS the the the user plane of the 5G core as well as the, the, the centralized and the distributed unit of the radios are deployed.

So, so far we covered uh the consumer business solutions, the customer journeys, the OSSBS platforms and the core and IMS systems that you deploy along the way. We, we understood generative AI capabilities that would be deployed. But there's one thing that you may ask about sustainability, that was the title that you read when you walked into this session.

Sustainability with Graviton three, with, with Graviton, what we have been able to do is now today this morning, Graviton four has been announced. So I have to be careful in terms of uh which version of Graviton chips that I talk about. But we have two, we have been able to achieve working with entity docomo uh a improvement of 72%. In terms of power efficiency, you can look out.

And, and I have limited time so i will jump in out of it but 72% power efficiency, we have been able to energy efficiency, we have been able to achieve. And in this space, we have been working on generative AI as well. There are three critical categories that we are working on closed loop orchestration, the ability to have analytics and and the ability to reduce the amount of manual intervention when adding capacity to the network is critical, close loop orchestration, the easier operations when it comes to root cause analysis, management, as well as providing in depth analysis and breakdown of KPIs and then user experience enhancement.

What does that mean that is going into the USS framework that Lana talked about and the ability to provide the user experience troubleshooting and user call simulations. On the same plane, all of this, if you're running a network that is uh in an on prem environment, not integrated into analytics and the, and the generative AI capabilities would not be easy to achieve. And these are things that we believe are critical for our customers as they evolve the 5G networks for both efficiency, as well as monetization.

The final piece is access networks. And within access networks, you understand, we we within those of you who are in the telecom domain, understand it's a highly competitive, very distributed space and it is very, very critical for customers to have a solution that scales across 10,000 plus locations. And this is a space where it is critical to understand how the architecture works.

So for those of you who are not close to this, the A has the RU the radio units which are then integrated into the distributed unit at the cell tower, integrating into the central unit a near real time R which allows for radio information controllers to be, to be handled. And then there is a non real time R which is the management function of the radio.

So when you translate that, how do I leverage AWS for these functions for the distributed unit uh that connects to the RU the radio unit, you can use Outpost and EKS anywhere uh uh as well. If it's, if it's existing infrastructure that you're leveraging for deploying the radio sites you have local zones where you can deploy your CU you can also deploy your CU on the Outpost based on the the transport capacity that you have available.

So how would that translate into an architecture? We saw this before? And i said i will come back to this. It is critical now to dive into the, the right uh left side for you of this architecture. We have the COS infrastructure if you see where somebody can deploy EKS anywhere, deploy virtualized D on top of it virtualized as well. And you also have Outpost servers which gives you full control extensibility from the AWS region, just the cloud at iran where you can deploy across thousands and thousands of locations.

Now, if you think about how would you go and deploy one of these for ORAN for and deployment is this is uh the framework, the same centralized observable, the same centralized analytics, all of the the pieces that you need starting from uh left to right. If you see in the region, you are seeing uh the, the non real time components, the management components that you need to deploy, then you have the regional sites where there is workload can be deployed across the CU en the CU the centralized units of the radios can be deployed on the gover environment here.

You also have the same benefits that i talked about the same CI/CD pipeline, the segmentation of network, the ability to uh deploy across parts of the network, have different level of observable and, and SMO capabilities and then have the distributor unit on the EKS anywhere environment on a quartz hardware as or or Outpost service.

So to sum it up, if you take the benefit of the virtualization capabilities available on AWS, we have worked on this for the last five years. A lot of things have moved in this space. The ability for our customers to differentiate with the network that is truly software defined. We have done studies with our customers is to achieve significant TCO benefits as well as opex reduction. 50 plus percent, we believe in terms of TCO benefits.

And we have as a matter of fact deployed CUDU sites very, very quickly on EKS anywhere these are available for you. If you walk up to the, the Venetian expo, you can see this entire network, not just the core but the radio access virtualization uh available for you. Um and we can walk you through how we have containerized and deployed this with our ISV partners across multiple ISV partners as well.

So that's what we wanted to walk you through in the last minute. There are two requests. Please do walk up to the team. They've put in a lot of effort to set up these live demonstrations that are in production at our customers globally. It is at Venetian expo both 580 to bring this again back to you next year, please do take the time to fill in the survey and uh and hope you enjoy it. Thank you for taking the time. Really appreciate it.