L4 Autonomous Networks approach first commercial deployments

To embed our video on your website copy and paste the code below:

<iframe src="https://www.youtube.com/embed/PUc3lK-k0Bw?modestbranding=1&rel=0" width="970" height="546" frameborder="0" scrolling="auto" allowfullscreen></iframe>
Sean McManus, TelecomTV (00:13):
Hello and welcome to WinWin Live. I'm Sean McManus, editorial contributor at TelecomTV. Today I'm talking to River Huang, CTO of Huawei General Development Department and Aaron Boasman-Patel, VP Innovation of TM Forum. They're here to share their experiences and insights on advancing technological innovations and accelerating the commercialization of Autonomous Networks level four. This time we have the special guest with us. Hi, WinWin. WinWin is an AI agent who will join our conversation today and bring us some unique insights. Mr. Huang, over the past year, the Autonomous Networks industry has developed rapidly and Huawei has been a key player in this. What is your outlook for 2025

River Huang, Huawei (00:56):
Last year the AN industry released level-four blueprint and the white paper version six. In the past six years, more and more CSPs published AN strategy and started the practice. Many achievements have been made to improve operational efficiency, user experience, and resource efficiency. AN has become the best choice for global CSPs towards network automation. Huawei has been an active player in AN industry. Over the past three years, we keep on researching on key technologies for Huawei ADN solution, contributing to industry standards and validating solution with our customers. Nowadays driven by 5g advanced deployment, digital business growth and transformation by GenAI, telecom infrastructure is facing big network operation challenges AN level four will be the key. In our opinion, 2025 will be the beginning of AN level four commercial deployment and it'll be a long term process including two phases: single domain autonomy in high value scenarios and multi domain, multi-agent collaboration in complex environment.

Sean McManus, TelecomTV (02:11):
Following so many years of development, it must be exciting to be reaching those first commercial deployments. TM Forum has built an important platform for the development of the AN industry. Could you share how TMF plans to continue advancing the AN industry in the future?

Aaron Boasman-Patel, TM Forum (02:26):
Absolutely. We've set up the AN mission board and we're working to expand the A manifesto to lead and drive the AM industry development, commercialization and promotion focusing really important on those high value scenarios. We are driving the commercial use and promotion of level four standards development are really critical for doing this, and we're collaborating with our industry partners to refine the relevant AN level four standards, such as the level four target architecture solution packages. So this is really key to driving industry forward.

Sean McManus, TelecomTV (02:57):
Achieving level four comes with many challenges and there's still a lot of work to be done. As CTO, Could you share your thoughts on the key technologies that need to be addressed?

River Huang, Huawei (03:07):
The breakthroughs in three key technologies are important to achieve level four. Large model, Agent and Digital twin. The first one is AI large model. Huawei provides Telecom Foundation Model for our ADN solution. We are developing copilots for different roles which can help to check network status, answer questions, and fix network faults. This improved operation efficiency and reduces the need for specialized skills. Only one large model can handle multiple scenarios, provide comprehensive solutions for both application and systems, thereby solving the problem of fragmented AI capabilities. At the same time, we see the large model technologies are developing very quickly, such as Deepseek. Huawei actively embraces them and continues to improve our intelligence capabilities. The second one is Agent. We are developing agents for high-value scenarios. These agents can realize network awareness, reasoning, decision-making, and self-learning, to achieve few or no human operation and create a new level for operation mode.

(04:20):
In order to build such agents, it's necessary to break through related technologies such as automatic planning and dynamic COT capabilities for unknown and complex scenarios, online experience, import capabilities and future-oriented multi-agent collaboration capabilities, including autonomous team interconnection, distributed efficiency communication, and distributed dynamic collaborative control and so on. The third one is about digital twin. As you know, we are building a holographic digital twin network. To realize this target, we should build a multi-modal space-time model based on kinds of network data such as topology, log, sequence-based KPI, packet and so on. This model can realize network convergence awareness and provide more accurate, real-time network simulation, fault prediction and prevention capabilities. Not only that, we should absorb the idea of "world models" in the field of intelligent driving to generate a massive amount of high precision network data to continue to improve the models generalization capabilities.

Sean McManus, TelecomTV (05:38):
Great. Across the large model agents and digital twin, it's clear making a lot of progress. Let's dig a bit deeper into the technologies. Large model technologies and agents are hot topics right now. How does TM Forum see the role of these technologies in enabling AN L-four?

Aaron Boasman-Patel, TM Forum (05:55):
Well, last year we jointly released the AN level four industry blueprint where we introduced gen AI into an target framework for the very first time. We've also established a general AI working group, leveraging the collective power of industry partners to continuously improve those related technologies and standards through innovation projects like the Catalyst Program, we're collaborating with our partners to explore and refine this application of those new technologies.

Sean McManus, TelecomTV (06:20):
It'll be exciting to see what new opportunities generative AI brings to the autonomous network. Huawei has been dedicated to advancing the key technologies needed for AN L-four and promoting their application. Could you tell us about some of your collaborations with Global Operators?

River Huang, Huawei (06:35):
We are working with our customers on level four high-value scenarios innovation. We have deployed a copilots and agents with China Mobile in Guangdong and Zhejiang Province and integrated them into daily operation to achieve automation with little human involvement, bringing huge improvement in operation efficiency. We are also working with other global leading CSPs, like Vodafone, Singtel and MTN to conduct a proof of concept tests. In the future, we are glad to work close with global industrial partners to help CSPs achieve AN Level-four targets.

Sean McManus, TelecomTV (07:14):
It's a great roster of operators to be working with. Now WinWin has prepared a special ending to wrap up today's discussion. Over to you, WinWin. Thank you all for joining us here at WinWin Live.

Please note that video transcripts are provided for reference only – content may vary from the published video or contain inaccuracies.

Aaron Boasman-Patel, TM Forum & River Huang, Huawei General Development Department

In a recent WinWin Live session, Sean McManus from TelecomTV discusses with River Huang, CTO at Huawei's general development department, and Aaron Boasman-Patel of TM Forum, the rapid progress and commercial deployments of Level 4 autonomous networks by 2025. They explore key technological advancements, the role of AI, and collaborations with global operators to enhance network operation efficiency, user experience and resource efficiency.

Featuring:

  • Aaron Boasman-Patel, VP Innovation, TM Forum
  • River Huang, CTO, General Development Department, Huawei

Recorded March 2025

 

Email Newsletters

Sign up to receive TelecomTV's top news and videos, plus exclusive subscriber-only content direct to your inbox.