Ray Le Maistre, TelecomTV (00:04):
Okay, we're in London. We're at future net World 2025. I'm here with Shahryar Khan, who's director and head of Mobile packet Core at the Telia Group. Shahryar, thanks very much for joining us at telecom tv today, at this event, a lot of talk about the use of ai, about network automation and autonomous networks, and let's chat about autonomous networks because that's something that's been talked about for a while, but it's really starting to come to fruition now. But can you just say from your perspective and from the market perspective, what an autonomous network actually looks like? What are the implications for this? Is it just the closed loop operations that people have been talking about for some years?

Shahryar Khan, Telia Group (00:57):
I think that the closest analogy you can think about from the autonomous networks is these self-driving cars.

(01:06):
So I think that's the easiest analogy that you have these cars. So I think the difference is that of course there are so many components that are automated and then they come together. But I think the difference from automation and autonomous in this case is that it is able to independently think about, for example, suddenly someone comes in the front so it needs to react or some obstacle comes in or suddenly someone. So I think it really has to be reactive in terms of something that just comes out of the blue, right? Independent thinking, intelligent thinking built into the networks. It's not just about having declarative statements or policies. If this happens, do this, this happens, do this. Of course that is there,

(01:56):
But on top of that, it needs to be able to react or think independently to come out of the situation. I think that is the difference actually for me, the key. And just like these cars, I think we have been hearing about it. I know for sure since a long time. And I think Tesla has, and a lot of others have done a lot of tests and they're there actually we have self-driving cars. But the thing is that the same thing can happen to the networks as well from my perspective. And it's not a matter of if it should happen, it will happen

(02:37):
To a certain extent. It's already happening. Everyone is in a different journey as you were discussing prior to this, to operators try to get there. Everyone has different, how to say it, reasons to actually take that journey. And some of them are ahead, some of them are a little bit behind. The journey is the same. All of them or us are taking different paths, but they all converge at some point of time. And for me, I think of course there's a lot of talk about this savings and there's a big business case if you do this right for me, that is if you keep that apart, fine, there will be gains, efficiency gains and you will be that aside. That is one thing, definitely it'll come. But for me, the key thing is this. The networks are becoming so complex. Maybe network is not becoming complex, but the applications and the way the use cases, they're becoming very complex and things are changing very, very fast. As we were discussing in this contrast, right, last year, no one knew about, no one was talking about agent ai.

Ray Le Maistre, TelecomTV (03:49):
Yes.

Shahryar Khan, Telia Group (03:50):
Suddenly there's all talk is about agents and this agent ai. So next year it could be something different. So that is the pace is very fast. So it's very important I think from my perspective, and operations of course are becoming very tough if I just look into it from the operational perspective, can we even operate this kind of a network? Because humanly I think it'll not be possible at a certain level. Even now, I think someone was saying in this conference earlier that there are millions and millions of parameters and KPIs and different layers, horizontal, vertical, which constitute a service. Anything can fail over here or here. It's very hard to figure it out where the problem is. And normally today of course people try to do manually. Experts are there, and of course there's some automation, but I think we are beyond that. We will be very soon, if not even now, beyond that point where it'll be extremely hard to even come out of that situation.

(04:52):
So I think it is a fundamental, from my perspective, a need. And also I think if just for me also, it's very important that everyone was talking last year about network APIs and exposure. So that's another aspect to it, that if you have this foundation of autonomous networks, then you can build anything on top and then it'll be very, very simpler to introduce new things like, so exposure on its own means nothing. If your network is not programmable, it's not autonomous, you cannot get the most benefit out of the network. API use cases, I can give you quality of demand, but if you cannot self adjust the optimize it right so that you can ensure that SLA that was guaranteed, it's going to be. And the thing is this, every use case will be different for us to be able to scale.

Ray Le Maistre, TelecomTV (05:51):
Absolutely. And that's when the developers will get really interested when those outside the current parameters are capable. So you mentioned there the progress that's being made and the complexity. So what are the foundational requirements for traditional telecom network operators to be able take this next step towards autonomous operations? So

Shahryar Khan, Telia Group (06:18):
I think again, in this conference, I think we heard a lot of, I think if I can select one thing, one key takeaway, which is common

(06:29):
Was having the right data. Data was mentioned many times that that's our Achilles heel, that's our pain point, clean data. But the thing is this data of course is important I think, and that's foundational, but you also need to put it into perspective. So from my perspective, I think if I take a very holistic view, I see there are three dimensions which constitute the foundations of building autonomous network. One of course is technology related. And data, let's suppose is part of that part. And then you have, when it comes to technology, you have different technology domains, right? In our networks like mobile packet code is one, IP transport is one, cloud is one, RAN is one. For example, OSS is one, BSS is one. So there are multiple silos still today. So the challenge is that, and data means differently when it comes to fulfillment flow data means differently when it comes to assurance flow.

(07:37):
And then for the closed loop it means totally different. But at the bottom, I think there is what the key is that you can develop as much common systems as possible. You should have one common inventory system, you should have one common way of collecting data. You should have one common way of building the cloud so that you can automate the infrastructure as much as possible, otherwise you'll not scale. So I think that that can tame the complexity a little bit of the overlying application that's going to hit us. So that is technology part. But other than that, I think what's important is the strategy, strategy and governance.

(08:17):
As you also saw, I think we had some couple of use cases where they had a program. Of course it was one use case, but there was a program that was run and there was a clear strategy that, okay guys, this is the value we are after, let's break it down into the steps and let's make it happen. Because TM forum automation levels are there. And someone also said that they're just the badges. But the main thing is that everyone needs to see that what is the value that you want to get out of it?

Ray Le Maistre, TelecomTV (08:49):
Absolutely.

Shahryar Khan, Telia Group (08:49):
Right? And your objective should be value-based. And then of course you can define journey, but it's very important that you define the strategy right from the start. And it should be aligned across all the organizations. Yes, everybody needs to know they, everybody needs to know that this is our goal for 2030, this is our 2030 vision, this is our 2028. And then break it down. Of course there will be smaller steps. And the other key part of the strategy is this, the legacy,

Ray Le Maistre, TelecomTV (09:18):
Yes,

Shahryar Khan, Telia Group (09:19):
Legacy is always going to be there. In some cases for some of us it is more. For some of others it's less. So for us, I think it's, in our case, I think it's very important for us that whatever we do now in the lifecycle, like cloud native five Gs A comes in, it's truly cloud native. So we should get all the benefits right from the day one. So there's no excuse, there's no legacy. So it has to fit in into the new architecture.

(09:49):
That's our I think, strategy. And then of course, you'll always have legacy, but you need to see that somehow both of them have to play together. Maybe you cannot get the full benefit out of it at point A when you have fully cloud native five Gs, which supports all these bells and whistles which are needed for autonomous city. But then you will always have legacy. Maybe in the BSS stat, there's something is in the SS stat, but still you can get some value. And I think the focus should be on that value and not get bogged down with the legacy. And then if I think philosophically, if I had a magic wand, I can just get removed my legacy at once when Rakuten came in. So if I can just remove everything and I can say that, okay, all my infrastructure is supports whatever is needed for autonomous building, autonomous networks, if you don't have the right strategy, because they're different moving parts, right? Every function, every network function, every application, even every microservice within a network function, it could have a different lifecycle management. There are compatibility issues, interop issues, vendors are not compatible

(11:11):
Even though we have standards and everything. So no matter if I do, even I do that, I will still even with a clean slate. Even with a clean slate. So it's very important to get your strategy. And of course you will fail, you can fail. It's always a chance. You'll fail, but you can adapt. So strategy should be flexible. So that's another thing. And governance, of course, you need to go on it very well like a program and make sure everyone understands this in the organization. And the last thing, which is very close to my heart, which I feel is the key to it all is the people side or the organization, right? Today, I myself, I was very deep IP engineer. I remember in those days, if you didn't ies, were always there. But if you're not working on the CLI, you were not considered as an expert. So we used to go deep down into the CLI hidden commands to do the debugs that no one else knew. And that was what we knew, that we knew the code inside out. But gone are the days that is not relevant. You have the machines, they can do much better job than us. But still the context awareness is still needed.

Ray Le Maistre, TelecomTV (12:23):
Absolutely, yes. Understanding how it all fits together and

Shahryar Khan, Telia Group (12:26):
Fits together.

(12:27):
So organization, I think we need to break down the silos of the organizations in the operator, maybe have more rather than domain centric teams. It should be more cross-functional teams, which are focused on the infrastructure, common components based cross-functional teams. And of course, you still need domain knowledge. I still need packet core experts who understand. I still need OSS experts who understand the orchestration. I still need assurance experts who understand that. So that is still needed, but then you need to have a T-shaped competence depth in that domain. But of course, you also need to have horizontal, broader in the cloud and ai. So my network, so if I go back, if I was starting today as an engineer, I would rather focus on those things. And of course the domain competence will come as well, of course. So I think this is the fundamental paradigm shift that I see from the organization perspective that needs to happen. Otherwise, I think we could fail as an industry.

Ray Le Maistre, TelecomTV (13:30):
But I do get the sense that, as you mentioned at the beginning, that everybody is on this same journey. I think the network operators and certainly the ones here, I think everybody understands that you can't sit and watch and wait and not do anything. And people on those different steps of the course, as you mentioned, everybody has different challenges and they're coming from different starting points. But I think there's a lot of commonality and which is why having these kind of conversations and sitting down as we will with some of your peers a bit later on, and having the broader conversation is actually very valuable to the industry and the audience. But you've given us a great encapsulation of what autonomous networks is about and really the foundations, which I think is absolutely fundamental and key to understanding this progress. So Shahryar, thanks very much for your time and for helping us to understand what this is all about. So thank you very much.

Shahryar Khan, Telia Group (14:34):
It was my pleasure. Thank you, Ray.