Tony Poulos, TelecomTV (00:05):
Hi, I am Tony Poulos and I am at TM Forum's, DTW Ignite Event in Copenhagen. A lot of talk about autonomous networks at this event and a little bit of talk about ai, but I have assembled a panel of experts here to help me understand autonomous networking and core network stability. And to help me out with that, firstly I'd like to introduce I Gusti Panca, who is the GM of research management and ICT, strategic alignment at Telkomsel. Welcome Panca. Good to have you here. Secondly, Andy Tiller from the TM Forum. Who is involved in this day-to-Day and I'm sure will be a wealth of knowledge for us. On the panel is Juan Carlos Garcia, who is a senior technology and innovations advisor to the industry as a whole. And last but certainly not least, Fernando Camacho here who is the autonomous networks product director at Huawei. Welcome Fernando. Gentlemen, welcome. Let's kick off with the main question. Firstly, what is driving the move towards autonomous core networks? Andy, I might get you to start off with that one. Well,

Andy Tiller, TM Forum (01:10):
Everybody immediately thinks saving money, but the motivations actually run much deeper than that. So as the networks get more complex, just managing them, keeping them running efficiently, you need the machines to compliment what the humans are doing today just to make it tractable. Of course, what you also get out of this is improved customer experience and ultimately if we can use automation to keep the networks running smoothly and efficiently, automate service assurance, automate the delivery of new products and services, they become tractable things that previously would not have been tractable so you can actually launch new services, which otherwise wouldn't have been feasible. So it leads to growth ultimately.

Tony Poulos, TelecomTV (01:53):
Well, and Juan Carlos, can you add to that? What do you think are the key drivers?

Juan Carlos Garcia, Independent (01:56):
Well, I think it's very important question to try to understand the rational behind the Oracle development. And this is the case as well. In the case of this autonomy in the core network comes from the need to cope with the complexity of the new services and features that 5G is bringing premium services based on new functionality like the network exposure function and the network slicing that will bring great changes in the traffic patterns and the traffic behavior that will be very dynamic and will change depending on the location and the point in time in the network. And this is something that will require adaptation, very dynamic will be very difficult for humans to learn faster, how the traffic is changing in the network and adapt to these changes and project will need a lot of machine intelligence to track all these changes, to adapt the network and to be able to respond properly. So then the reason mainly for this is as Andy said, is to cope with this new functionality, sophisticated functionality that we're bringing.

Tony Poulos, TelecomTV (03:05):
Well I'd be interested to hear the network operators viewpoint Panca, what's your driver?

I Gusti Panca, Telkomsel (03:10):
The most important actually because we have a very lot subscriber is around 160 million. And then with those subscriber, actually we have a challenge in terms of the operational, we have two kind of principle that are running our core and then we believe with the some sustain and then together with some AI inside through this autonomous, especially in the core, which is we have running our platform right now for the run autonomous network in Iran, even though until to level four, maybe 2025. So I think this is the best chance to follow to continue our e evolution journey. Not only stop in the run but also into the car.

Tony Poulos, TelecomTV (03:58):
Fernando, from a vendor's point of view, what are you hearing? What are you being requested and what are the drivers that you are seeing?

Fernando Camacho, Huawei (04:06):
So I fully agree with my colleagues the fact that 5G is a complex technology core network that is very, very complex as well. So really for the operators just driving the opex of driving these technologies is increasing so much. It's actually eating up the revenue, the revenue growth. So autonomous network really is the right, I guess, framework to bring that minimize the level of human intervention so we can get the system to do the work. And it's also a great framework to bring AI to bring LLM to bring gen AI into the network. So it's really the way forward to try to reduce, to bring network efficiency and reduce the opex that is hitting up these revenues from our colleagues operator.

Tony Poulos, TelecomTV (04:50):
So how do you assess autonomous core network levels from a global perspective? Andy, perhaps you could respond to that.

Andy Tiller, TM Forum (04:58):
Sure, yeah. Well a while ago TM forum members developed a very simple scheme for measuring the level of autonomy in a network. So from level one to level five, a lot of our operator members have set a target of level four by the end of 2025. We have a detailed scheme for doing this assessment and we've been running a pilot project with telecom cell involved. Fernando has been the co-chair of our project to do this where we've had 10 operators worldwide comparing their own level of autonomy using this scheme and we're creating a very standardized approach to this. So there'll be an industry benchmark.

Tony Poulos, TelecomTV (05:38):
Panca, I don't want to know what your score was, but I do want to know how you were using it and is it an advantage for you to use a system like this to assess the network?

I Gusti Panca, Telkomsel (05:47):
Well actually we use the tools that provided by TM forum when we assess since the beginning where we are in the beginning we are only in the 1.8 for our score. And then now because helping by our partner everything, then hopefully we can be reached into the level four, maybe by end of 2025, even though actually we have some struggling for the use case. But this is not only as an end of our journey because currently we don't have the 5G standalone yet, then it'll be some dragging when we are talking about network slicing so on, then absolutely we will require this kind opting for our will trial ency. So on

Tony Poulos, TelecomTV (06:37):
Fernando, you're being asked to help out to get the numbers up, get the levels up. Are they calling you in?

Fernando Camacho, Huawei (06:45):
Well obviously as Andy said we just run the pilot for what was very important to standardize these, the tools that operators use to measure what level they are. The thing here is that they're targeting level four, level four, maybe next year, I know 2025 or in a few more years. You need to know where you are in order to understand what you need to do in order to achieve that target. So it's very important to understand those weak points that you have today in order to improve those processes. Right. So these tools that we are trying to standardize after this pilot will help the operators do that, understand where they are in order to understand what the roadmap, what they need to do in order to reach the magic level four that they all want to be.

Tony Poulos, TelecomTV (07:23):
Well. And Juan Carlos, are you advising your clients to use the two?

Juan Carlos Garcia, Independent (07:27):
Yes. Yes, of course. As Fernando says, there is the need to somehow measure to understand how far you are from your target. But I will give you my general view on where we are. I think we are not just introducing a new technology in the network. So autonomy is not just about technology, it's also about changes in the way organizations work. Traditionally we have seen operations run by teams of specialized experts running each domain on their own with assisted by tools. We'll see over time how these humans have an artificial intelligence assistant or agent working with them and improving the productivity to give the next step. That is once they get sufficient trust on the artificial intelligence that is working with them, they will leave the artificial intelligence to work on its own to reach the level of autonomy. I think that today we are in the process of bringing artificial intelligence assistance to humans in the moment of start to gaining trust on the artificial intelligence to be able to give the next step that is leaving the machine, taking decision on its own, on my view. Also, at the end of this journey, you'll not have absence of humans. So humans will be there always setting the intents, setting the objectives, monitoring the learning process, monitoring the models that the artian generates to understand if they are rational or not, and also providing the touch, the touch that you need sometimes in a service that is provided to castings. Okay,

Tony Poulos, TelecomTV (09:00):
Stability comes up all the time. We're obsessed as an industry with stability and always connectedness, if I can use that terminology. How important is stability Now when we talk about autonomous core networks? Andy, did you want to kick that one off?

Andy Tiller, TM Forum (09:15):
Actually, maybe I'll defer to the guys who are actually really doing this in the real world. In the real world, yes. So maybe,

Tony Poulos, TelecomTV (09:21):
Yeah, Panca, tell us about how important is that stability factor?

I Gusti Panca, Telkomsel (09:25):
Yeah, like what I already explained, that we will have a standalone with the NetApp slicing play with the very low latency sensitivity in the network. Of course it will be like something that we need to be anticipated using this kind of tools, everything in terms of the stability itself because this is very important to make our customer comfort and then they don't feel that there is some dragging into their performance when they use the surfaces.

Tony Poulos, TelecomTV (10:05):
We're also talking about the relationship of these core networks with other networks works that are operating within the operator. How important is that interaction and stability between all the different systems and networks?

Fernando Camacho, Huawei (10:17):
I guess reliability of the networking itself is crucial. If you cannot trust your network really to get into a lot of trouble for us from the huawe point of view is very important to have the standards that will really help you measure that stability, really understand whether you have a stable network and help obviously our operators that they can measure that stability. So for us it's very important to bring this standardization into the stability area and they can have the right tools to measure the stability within the network. So with all technologies we core networks especially

Tony Poulos, TelecomTV (10:51):
Well Juan Carlos, it's very dangerous if work isn't stable.

Juan Carlos Garcia, Independent (10:54):
Yes, and I think stability is something you need to watch when you go to autonomous networks because at then you'll have a lot of artificial intelligence agents taking decisions in different domains and in some cases they may take contradictory decisions that can put in risk the stability of the network. Think about, for instance, an agent that is trying to improve the energy efficiency while another agent is looking at ensuring the quality for a certain customer. One is trying to change a configuration in one direction and the other one in the

Tony Poulos, TelecomTV (11:28):
Could affect the other direction.

Juan Carlos Garcia, Independent (11:29):
They can start doing continuous changes on the network and somehow you need the humans to be able to set the policies, to set the priorities, to be able to have the rules for solving those conflicts in this artificial intelligence network.

Tony Poulos, TelecomTV (11:46):
So Panca, what happens after you've done the pilot assessment and you realize the status of your network? What's the next step?

I Gusti Panca, Telkomsel (11:55):
Actually in telkomsel after we've done for our autonomous network, especially in the run, because that is really related to the quality performance. Then thanks to 10 forum that there is an initiative with the core, which is also, this is something that we are looking for and thank you so much, we can be here and then being organized by TM forum to see even further about what actually we can leverage more into our technology intercom sales, especially for the core. So I believe that this is something that we can continue further for our collaboration with the TM forum and I hope we can meet you in Jakarta later on to have a more discussion with our engineer team to see the prospect for this kind of solution.

Tony Poulos, TelecomTV (12:48):
So Andy, the TM forum and its colleagues and friends have developed this assessment tool, but what happens after that? What's the next stage in the process when they get to the magic for

Andy Tiller, TM Forum (12:58):
Right. So we've got the diagnosis today, maybe 2.5. The target is for of course that's going to cost money to get there. You need an automation project. So the first thing you need to understand is the return on investment. We have a set of tools that help with that key effectiveness indicators, which tell you what's the business benefit you expect to get from getting to level four. And then after that, of course you need to know how to get to level four. So we're developing in the autonomous networks project a set of blueprints and APIs and reference architectures which provide the guidelines for our members to actually achieve the level four automation that they're looking for.

Tony Poulos, TelecomTV (13:35):
And presumably, Fernando, you follow these rules with your clients and develop systems to support them?

Fernando Camacho, Huawei (13:42):
We do. We do. Obviously, as Andy said, the average now is 2.5. In 2.5 and three. The networks are still run by the human highly skilled engineers, but with assistance of some tools and for example, while we develop the right analytics tools that the help then bring to this level three, we order to get to the level four, you definitely need to use the ai. We are actually working with the gen AI to try to bring all the 30 years experience we have with core networks to try to have all of that embedded into the intelligence that we're going to bring to the level four. But there is a challenge to get to level four. We don't want to, it is not an easy task to get to their magic level forecast.

Tony Poulos, TelecomTV (14:23):
If it wasn't a challenge, we wouldn't bother with it. In the industry, you know that Juan Carlos ai, is it going to be critical in autonomous networks? Do you think it'll be an essential component?

Juan Carlos Garcia, Independent (14:34):
Well, just to give a different view to the topic, I think AI can even be used to track the progress towards autonomy. Because at the end, what you are looking for is to reduce the human intervention, especially in those tasks that are repeated, that routine and that the machine can do better, faster, and more often. And in this way you can always track how much the humans are using the systems or look at the logs and look at whatever, and then make a continuous assessment on how your networking general is evolving towards autonomy. And I think it would be very interesting also to do this kind of tracking, to really assess in real time, day by day, week by week, every month, how progressively humans are touching less than network and machines are taking and taking over some of the decisions over.

Tony Poulos, TelecomTV (15:26):
You make it sound so easy. I'm sure it's going to be a bigger challenge like Fernando was saying, than expecting Panca. Thank you very much for being with me today, Andy, Juan Carlos, and Fernando, thanks so much for your input and I think I gained a lot more knowledge now about what's happening with autonomous core networks, particularly with stability. Thank you again. Thank you very much. Thank you so much.