Leading the Way in Large Model Application Innovation and Advancing Toward AN Level 4
[Shenzhen, China, April 18, 2024] At the ADN Session of Huawei Analyst Summit 2024, Ken Lu, President of Huawei's General Development Department, delivered a speech highlighting the achievements in commercial Autonomous Networks (AN) Level 3 practices that Huawei and communication service providers (CSPs) have made together. He also introduced key technologies, such as the telecom foundation model, shared insights on the progress of innovative ADN Level 4 solutions based on high-value scenarios, and invited all industry partners to collaborate on achieving AN Level 4.
Ken Lu, President of Huawei's General Development Department, delivers a speech
Lu stated, "With the assistance of the telecom foundation model, the transition from Level 3 to Level 4 will achieve a quantum leap for value creation. Throughout this transition, all industry contributors need to reshape the O&M mode, system capabilities, business processes, and integration mode to enable CSPs worldwide to attain AN Level 4."
Huawei has introduced a telecom foundation model based on the AI Agent architecture. This model provides role-based copilots and scenario-specific agents to enhance employee productivity and improve customer satisfaction. Using the natural language processing capability of the model, copilots create customized digital assistants for different roles and carry out human-machine interactions using natural language. The model also features advanced intent understanding and logical inference capabilities, enabling agents to break down complex tasks and provide optimal solutions. Agents can also leverage system capabilities through APIs, achieving autonomy in specific scenarios.
The capabilities of copilots and agents enable the development of applications. Huawei collaborated with China Mobile Guangdong to launch the first application based on the telecom foundation model — Net Master. Net Master possesses tens of billions of telecom corpora covering all telecom Q&A scenarios, understands user intents, and performs fault self-diagnosis. The application simplifies complex and manual network configuration and troubleshooting, reducing the end-to-end (E2E) fault rectification duration by 20%. For core network O&M, Huawei and China Mobile Zhejiang jointly introduced the first application based on a multi-modal foundation model — ICN Master. In monitoring, troubleshooting, and complaint-handling scenarios, ICN Master automates and streamlines key processes in alarm ticket and complaint ticket handling. The E2E complaint handling efficiency improves by over 50%, and the overall troubleshooting efficiency by 33%. The use of large models has proven to be effective, helping CSPs redefine networks from four aspects.
To expedite the innovation in large model applications and assist global CSPs in systematically achieving AN Level 4, Lu called for collaboration among industry partners to accelerate talent development, transform processes, and apply large models in high-value scenarios. At the ADN Session, industry analysts from the China Academy of Information and Communications Technology, China Mobile, China Unicom, and Accenture were invited to deliver speeches and share insights on the AN industry, advancements in large model technology, and best AN practices. They aim to advance the growth of the AN industry collaboratively.
