From Automation to Autonomy: How Security, Observability, and QoE Drive Next-Gen Networks

The future of connectivity is here, and it's autonomous! 🤖

‍

Host Will Townsend, Vice President & Principal Analyst - Networking & Security Practices at Moor Insights & Strategy, is joined by Boost Mobile's EVP and CTO Eben Albertyn and Nokia's Cody Bowman, VP, Solutions Consulting, for a conversation on AI and automation shaping the next generation of Wide Area Networks (WANs) and the telecom industry. The future of connectivity hinges on these strategic shifts: a move beyond theoretical autonomy to the practical implementation of self-driving networks!

‍

Key takeaways include:

‍

🔹Security Evolution for Autonomous Networks: Networks must undergo a fundamental security transformation to address expanding threat surfaces and enable full autonomy.

‍

🔹Data-Driven Network Intelligence: Real-time data correlation and observability are crucial for the machine learning that will drive predictive intelligence and network autonomy.

‍

🔹AI-Powered Customer Experience: AI automation has the potential to revolutionize customer interactions, enabling proactive issue resolution and personalized experiences.

‍

🔹Strategic Imperatives for Network Autonomy: Achieving truly autonomous networks requires investment in technology, talent development, and a flexible approach to evolving industry paradigms.

‍

Learn more at Nokia and Boost Mobile.

‍

Watch the full video at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.

‍

Or listen to the audio here: 

Will Townsend: Welcome to another Six Five Media Virtual Podcast. I'm Will Townsend. I manage the networking and security practices for Moor Insights and Strategy. And joining me today are Nokia's VP of Solutions Consulting, Cody Bowman and Boost Mobile CTO Eben Albertyn. We're talking about autonomous networks and the role of AI and automation in shaping next generation connectivity. Welcome gentlemen to the conversation.

‍

Cody Bowman: Thanks Will.

‍

Eben Albertyn: Thank you, Will.

‍

Will Townsend: Yeah, great to have you here. So Cody, I'd love to start off with you. And let's talk about security. And it's often the biggest barrier to automation and mission critical networks and it's based on the disaggregation of infrastructure, expanding threat surfaces. So I'm wondering if you could share your thoughts on how security architectures can evolve to enable full autonomy while ensuring resilience against emerging threats.

‍

Cody Bowman: So Will I think one of the key areas in security, especially as we move forward where telecoms networks and IT networks are more and more the line is more blurred all the time as we get into the evolution of how networks are built these days in the telecom space and Eben and of course can talk to that is they have one of the most advanced networks in at Boost Mobile than in everyone else in the industry because of the way they've designed it. And part of that is really a kind of a seamless design approach where you end up with things like identity access management and endpoint security, SIEM and other qualities of security capabilities that are so important and integral to being able to properly secure any telecoms network end to end and any IT network that manages pieces of that or OSS BSS as it's traditionally called. But the real key thing is to be able to get that proper design from the very beginning. And a lot of, you know, brownfield evolution networks, it's a challenge. It's something that is a continuous, growing and evolving play. But even for Greenfields, it is as well too. So I think that one fundamental area is to recognize that there's fundamental differences between IT needs and telecom systems needs. And being able to find that balance is critical. The second thing I would say is that more and more we're seeing real demand for disaster recovery solutions. And those solutions basically are actually something that go well beyond individual components and redundancy and geographies and typical things that we've done in the past to make security something that could be distributed. It now is something that we can actually stand up completely separate networks in real time that replicate the original network. And we demonstrated this at Mobile World Congress recently and it is something that resonated very well with our customers worldwide. So I think that the ability to have disaster recovery resilience is key as well.

‍

Will Townsend: Yeah, I would agree. And Eben network observability data observability is pretty hot these days. It provides network assurance. It also can bolster security posture as well. And as networks become more autonomous, that's going to be a very important aspect of ensuring proactive management. So I'm wondering from your perspective at Boost, what's the next frontier in predictive intelligence and observability?

‍

Eben Albertyn: So near real time or real time correlation of cross domain data sets that then also feed observability creates the data ecosystem that one needs for machine learning. Now with properly trained models one can then move on later to use AI and then you have a chance at progressing at insights. And so keeping data integrity very high is absolutely crucial. I'm sure you've heard everybody pretty much start and finish their conversations with quality data. And then from these models that have the ability to comprehend causality, only then do you start to attempt working towards supervised activity that these agentic models will then attempt. The next frontier would then be agents that are able to interact with each other, that understand causality and that has a reliable ability to affect change and then ultimately has the ability to pursue a set of complex goals and or react to other agents. That for me would be the next frontier that we are aiming at achieving with the network that we've put in place. And then you know, some of the challenges that we face in getting there is conflict. So agents end up trying to combat each other in terms of the outcomes they're trying to achieve. Drift and hallucinations proves to be a continuous area that we need to focus on. And as I said, data quality.

‍

Will Townsend: From my perspective, a lot of the focus around generative and agentic AI within the telecom space has been in kind of reimagining customer support and that sort of thing. And quality of experience is a really important aspect of what you deliver, right? From a telecom services perspective, when you provide an excellent quality of experience, you reduce, churn, you create stickiness with your subscribers and that sort of thing. And it can be a really strategic differentiator. So I'm wondering if you could spend a little bit of time and talk about how AI driven automation can drive that seamless customer experience and adapt the network to become truly sort of self driving or self healing because there's been a lot of discussion about that over the years, but we haven't really seen it. 

Eben Albertyn: So telecom networks and also consumer interactions, if you were to think about OSS and BSS, are fairly rule based. Depending on what you compare it with, it could be even very rule based. And hence, at least in principle, it should lend itself to very, very successful ML model training. This potential that resides inside the data that constitutes telecom networks then opens the possibility for huge amounts of data to be correlated with each other more than what we can today. And in addition to that with intrinsic knowledge of the system as a result of machine learning. Because we've created a model that understands causality. This in turn will help us to tune, optimize, troubleshoot and resolve and eventually prevent issues that customers experience because of the data not only being available, which it's been for a long time now. Because a tremendous amount of intelligence can be applied across a much larger data set in a much shorter amount of time. And that helps us to be able to then achieve and start to head in the direction of getting to a place where we can actually resolve customers' challenges. We see that in practice starting to bear fruit, especially in the network related area and with network related activities. I think areas that most of us probably have experienced some frustration around this area, for example in the area of customer care, chatbots for example, I think pretty much everybody would recognize they don't really work yet because they do not interact like a human, they hardly ever understand what you're asking them. But the network structures underlying are very, very rule based. And so those ML models tend to be able to learn them and understand the causality that happens there quickly, quicker. And so we are seeing the benefits coming through there definitely very, very tangibly. And then I'm talking about months, not, not, not years or something.

‍

Will Townsend: Yeah, I mean we've all had that frustration, right? You know, we try to get an answer to something, we can't get that answer, we call a phone number, we're on hold forever and it just becomes just a very, very poor customer experience. So I mean this is a huge area where I think AI, you know, modern AI can really make an impact, but the shift to autonomous networks is a pretty heavy lift. Cody, I've written about this. Nokia commissioned a paper through my firm Moor Insights and Strategy. I've written some contributed writing to our website as well, talking about some of the things that we've discussed around observability and that sort of thing. But again this is a sea change approach when you look at Nokia's definition of the autonomous network. So I'd love Cody to start with you. What are some key challenges that organizations are facing here and how can they begin this journey?

‍

Cody Bowman: Well, Will, I think, adding on to what Eben was talking about from a QOE perspective at a consumer level, there's also a QOE that we're looking at from an enterprise level, from a variety of different verticals. I mean, you can almost pick any of your industries that the CSP community is targeting, if you will, for exposing network assets to them, to give them value like QOD and precise location and identity management and controls. And that adds a brand new dimension, if you will, of operational autonomy that networks, CSP networks in the telecom space need to be able to respond to. So now you've got conversational autonomy that's going on between the IT networks in the enterprise space and then the telecom space, being able to respond to the request of the IT networks and being to then respond and restore the capabilities of the telecoms network back to its original state once a request has been fulfilled. So that conversational network autonomy is something that is a new level beyond just the business model of programmability. It gets into the ability of the telecoms networks to act and walk and talk differently and behave completely differently. So the workload that is designed formally for more of a human interaction, either from enterprise or from consumer applications, is now starting to really change and become more about machine to machine at a network level, communicating together.

‍

Will Townsend: Now that's a great point. And I liked how you contrasted the difference between enterprise networks and telecom networks. I mean, the number of devices on telecom networks are massive. Right, Eben, and so what are your additional thoughts on this point?

‍

Eben Albertyn: So the key areas of focus initially I would very much start with what level of autonomy and where in the organization or where in the customer journey do you want to achieve that? I think we continue to have very robust conversations about what we define as the level of autonomy that we see that we need to be able to have to be effective and to be safe. I think first and foremost, the next topic that we focused on is data security and data management. We've touched on this before, but this includes the initial elements of things like data retention, security, tagging, but then also eventually moving forward to data quality as well. And if you eventually, as Cody alluded to, have machines starting to communicate with machines, you may have machines creating data and then that data will be reused in another place. So monitoring that data quality in there, especially if machines are creating new sets of data as a result of their interactions. That certainly is very important to us. You started with security, but that's also very important for us are the security frameworks in which we are actually going to operate. And that's very relevant in terms of how compute environments are migrating and increasing. Automation by itself must not be able to reduce your security posture. Automation must comply with and further the security framework that you've built on an active basis and so you know things to take into account that certainly are important for us is that certainly the fundamentals need to be in place, you need to understand the risks. Zero trust environments, which is what we've implemented, provides you with a role based segmentation ability to access your networks. And there are some other considerations around security, but those certainly are very important to us. Then people training, skills development and governance. And not governance in the terms of what are the rules of what you shouldn't do, but more around the rules of how we can help each other, how we can be effective, and how we can share outcomes with each other so that when one group gains, another group also wins. As we move forward touched upon drift and hallucinations, that remains to be an area that we cautiously look at constantly in terms of being able to understand. As models get smarter and are able to understand more, are we able to understand what they are actually objectifying and looking at. Orchestration architecture is becoming very important. Like I said initially, the concept is that the models are able to understand causality. The next step would be for models to be able to affect change effectively. Large language models do that today for us. They can speak, they can communicate with us in the same way. These models will interact with systems and give them instructions. And then lastly, you want the ability to be able to pursue a set of complex goals. And then these three will then become a symbiotic chain of events where that will continue to happen and how that architecture is then orchestrated is crucial to how you eventually will then attain autonomy, first in a smaller part of your ecosystem and then eventually start to grow. And then API frameworks and API security is another area that we're very much looking at as we want agents to be able to speak to agents. Because we see the large language model type interaction is probably less than 10% of what we'll actually do with AI. It's more about how they will communicate to themselves. And then best practices. I mean it's been used for a long time, but crawl, walk, run, it's a great way to get things moving. Like I said, have a clarity around the vision that you have for the future. Where do you want to have autonomy? What is your definition of autonomy and what is the definition of success in the area you want to move towards. Clarity of use cases, one gets confronted with a whole lot of alternatives, options, possibilities. But it's important that you understand what use cases you actually want to pursue and why so that you also understand when you actually get where you are. And then I think it was Stalin that actually said this, but trust is good, control is better. Or maybe a German saying, I don't know, but that's even with robots. So I've touched on hallucinations, but one needs to make sure that the underlying framework that you have has an ability to not only report support what they are doing, but also that we, we have a mechanism to be able to understand what's actually happening and if that's still congruent with what we're trying to do. So those are kind of some of the best practices that we think would be super valuable to take on board or at least consider.

‍

Will Townsend: Well, that's a great perspective. I appreciate you sharing that. And, the potential for autonomous networks to, to transform the way we communicate is huge. The application of modern AI we've talked about, agentic, generative AI is tremendous. I mean all of this, it comes down to automation. You have to have high degrees of automation. But I'm wondering as we close out our conversation, Cody, let's start with you. What should CSPs be thinking about? What should even enterprises be thinking about that are considering private cellular networking and autonomous networks? But what key investments need to be made up front so that these networks can be future proofed?

‍

Cody Bowman: Well, clearly, as Eben pointed out, the nature of data is foundational to ensure that everything is going to be completely connected well through the end to end process. As things evolve, as people evolve, as the automation continues to improve and large language models, agency AI and all of that ecosystem begins to be more coherent, then it will start to prove itself. We're going to start to discover things that we don't know now because we're making a lot of, you know, very educated guesses based on what we do know now. But time will tell about how it actually plays out. And then we need to make the adjustments appropriately. Especially when it comes to security, especially when it comes to zero everything is what I refer to sometimes zero trust, right? Zero weight, zero trouble. I mean there's this ambition level that I think industry has to try to get the automation to a point where the human is the observer of the network and looking at the values that it brings and being able to make modifications in real time based on intent, based on service monitoring and service observability instead of network monitoring and observability. These are all going to be fundamental shifts. And I think in that space, the investments that the industry is making in this area has to go not only into the technology, but also into its people. As Eben mentioned earlier, that is so key because the cultural shift that will happen in the workforce and the way people behave, the way they interact with the networks will change pretty dramatically. And so I think that and the way that the sales organizations within all of our industries operate, especially the way that the business looks at the go to market motion, is going to change dramatically as well too. Today managed services are the norm in the business. Tomorrow its people come in from all walks of life, all enterprise types and access telecoms assets to use for their needs. And the network then responds to it, it controls it, it manages it in a way that is profitable and secure, but provides the capability and the flexibility that these enterprises need. And that's, that's a tall order to get to that, but that's the direction we're heading.

‍

Will Townsend: Yeah, no, that makes perfect sense. Eben, your final thoughts on this subject?

‍

Eben Albertyn: Yeah. So in terms of investing for what will be worthwhile to capitalize on as soon as one starts to go on this journey, top of the list would be people and skills. They are going to be the means to how one reaches the objectives that you want to reach. Second would be clean data, secure data. That would definitely be a worthwhile area to invest in. I think Cody alluded to it. But the first thing that's really worthwhile investing in, and we can see that on a daily basis and invest with an open mind. There are paradigms which we have convinced ourselves in the past are important. And potentially in the past they were important, but potentially things have changed quite a bit. And as things start to progress and how we move forward with AI, even the way that we thought about using ML&AI 18 months ago may not be the best way to do that now. And so I would say people clean data and the benefit of an open mind would be great.

‍

Will Townsend: Yeah, great point. I mean it's still very early days when you look at modern AI. I mean I, you know, I have to educate folks that, you know, journalists and others that I talk to that, you know, AI and machine learning have been around for decades, but it's this notion of modern AI that is so new and the landscape continues to change. Security is important. Runtime security is important. Telecommunication networks are huge potential targets by bad actors that want to weaponize the same modern AI that is beginning to transform, you know, injuries industries, including, you know, telecom and retail and in manufacturing and whatnot. But, gentlemen, it's been a very, very compelling conversation. I appreciate all of your insights and I want to thank our viewers for tuning in to another Six Five Media Virtual Podcast.

‍