Cross-Domain Synergy: ZTE and Guangdong Mobile Build an Autonomous Network Demonstration Zone

With the rapid development of the digital economy, 5G, gigabit optical networks, cloud computing, and industry-integrated applications continue to evolve, while network scale expands and service models become increasingly complex. As a leading Chinese communications operator, China Mobile Guangdong (Guangdong Mobile) is facing unprecedented challenges.

In 2025, Guangdong Mobile and ZTE jointly launched the "Autonomous Network Demonstration Zone" project, promoting the transition of network O&M from "passive response" to "active prevention" and accelerating the evolution from L3 to L4 autonomous networks.

A Full-Stack Intelligent Evolution Path

To address common pain points in network intelligence transformation, including data silos, fragmented intelligence, and broken loops, ZTE and Guangdong Mobile have adopted a cooperation philosophy of joint research, scenario co-building, and ecosystem collaboration to develop systematic solutions.

Leveraging its self-developed Nebula Telecom Large Model and full-stack intelligent solutions, ZTE effectively breaks down technical barriers and promotes the evolution of network O&M toward high-level autonomy. Anchored by a “1+N” strategy—where “1” represents the unified telecom large language model, and “N” refers to diverse auxiliary models, including structured models, graph models, digital twin models, and other industry-specific models—the system employs a three-tier architecture comprising the network element, single-domain, and cross-domain layers. This framework integrates Copilot-assisted O&M with agent-based automated handling to achieve an end-to-end intelligent closed loop. Building on this foundation, an intelligent decision engine is introduced, featuring foundation-model flexibility, task autonomy, dynamic knowledge, and rigorous reasoning, driving the network’s transition from pre-defined rules to autonomous cognition.

The two parties focus on four core domains— wireless network, core network, transmission network, and network cloud—and integrate AI agent technologies into network O&M scenarios, significantly improving O&M quality and efficiency.

Wireless Network: Building the Industry's First General Fault Expert Agent

To solve the issues of traditional wireless troubleshooting, such as heavy reliance on manual experience, time-consuming fault delimitation, and low coordination efficiency, ZTE and Guangdong Mobile jointly launched the industry's first general fault expert agent. Deeply integrated with Guangdong Mobile's "micro-grid" production management system, the agent reconstructs the wireless maintenance process.    

The project builds a unified perception layer covering multiple sources of data such as alarms and KPIs to achieve panoramic fault visualization. It simulates expert diagnosis logic to automate the full process of "abnormality identification–root cause location–handling recommendation.” By deploying intelligent agents, the system can automatically generate work orders, trigger parameter adjustment, and perform remote recovery operations, forming a closed-loop process.

As a result, the accuracy of wireless fault root cause diagnosis increased to over 90%, the accuracy of knowledge Q&A increased to over 90%, and MTTR was reduced by 21.34%. The project received recognition as a TM Forum Innovation Hub Pioneer Project 2025 for “Multi-Agent Collaboration: Driving E2E Closed-Loop RAN Faults Management.” 

Core Network: Building an AI-Agent-Based Automated Fault Handling System

As the service center, the core network has extremely high requirements for stability. However, fault causes are complicated, and traditional monitoring methods cannot respond in a timely manner.

The project builds a core network fault agent, establishes a cross-NE correlation analysis model, and integrates fault knowledge graphs with deep learning algorithms. This improves the timeliness of fault risk identification, provides fault delimiting conclusions and recommended actions, and enhances efficiency of closed-loop work order processing. For the first time in Guangdong, an innovative one-click handling capability has been introduced and workbench integration has been completed, enabling one-click operations such as isolation, switchover, and restart.

In this project, both the coverage rate and accuracy rate of intelligent fault diagnosis in the core network exceeded 90%, realizing intelligent and efficient fault handling.

Transmission Network: Driving Energy-Efficient and Low-Carbon SPN Operation

As a key bearer platform for government-enterprise private lines and cloud-network integration, the SPN network features high device density and high power consumption, making energy efficiency optimization an urgent task.

By combining AI capabilities deployed at the network element level with centralized AI capabilities at the network level, dual-layer intelligent management is enabled to automatically formulate optimal energy-saving strategies, achieve endogenous intelligence, and support capability exposure.

Based on the core architecture of "dual-layer intelligence and end-to-end energy saving", the project adopts a hierarchical, collaborative design to build a full-link energy-saving ecosystem and a service-lossless policy engine. Using long- and short-term prediction algorithms, the solution analyzes network traffic in real time and dynamically adjusts the energy-saving status of devices, which enables full-process automated energy saving while supporting service assurance, intelligent energy-saving prediction, and energy consumption visibility and controllability. Power savings for a single device exceeded 15%, and annual average electricity costs can be reduced by more than RMB 1 million. A replicable green network operation template has been established.  

Network Cloud: CUCD Enables Efficient and End-to-End Change Management  

Due to the high frequency and high risk of network cloud changes, as well as the need for manual coordination, hidden faults caused by improper human operations must be strictly prevented.

By opening up OMC real-time data reporting capabilities and automatic network operation interfaces, the system is fully interconnected with the network cloud workbench. Centered on the three stages of "pre-event prevention, in-event control, and post-event verification", the system strengthens self-warning and self-isolation capabilities for change risks at key nodes, including joint review of change solutions, approval of work orders, task preparation, result testing and verification, and on-duty observation during cutover. This establishes an intelligent change monitoring system, enabling end-to-end change automation and ensuring that risks remain manageable and controllable.

In this project, the automation rate of the entire change process exceeded 90%, significantly reducing the risk of human errors, comprehensively improving the standardization, security, and efficiency of change operations, and providing powerful support for building a highly reliable and autonomous network O&M system.

In addition, the two parties completed the deployment of pilot sites for 12 high-value scenarios, including network cloud fault monitoring, transmission network fault monitoring, wireless performance optimization, wireless energy efficiency optimization, and transmission change monitoring, initially building a capability foundation for L4 autonomous networks.

From Technology Breakthroughs to Ecosystem Leadership

The Autonomous Network Demonstration Zone was built over one year and has achieved remarkable results. It has not only significantly improved key O&M indicators, but also established a replicable model.

ZTE and Guangdong Mobile jointly participated in the technical achievement verification of the "Co-Innovation+" Autonomous Network Lab and applied for the Science and Technology Award of the Guangdong Communications Association. Multiple achievements were showcased at the Digital Transformation World (DTW), PT Expo China, the China Mobile Partner Conference, and TM Forum Innovate Asia 2025 in Bangkok.

Based on the Autonomous Network Demonstration Hall, the two parties have built an open platform for achievement demonstration, technical exchange, and talent training, continuously empowering the industrial ecosystem.

This successful practice is a milestone in the in-depth collaboration and joint innovation between ZTE and Guangdong Mobile, and an important step toward higher-level autonomy in the communications industry. From pilot exploration to benchmark demonstration, and from partial optimization to global intelligence, ZTE will continue to work with Guangdong Mobile to develop the demonstration platform into a hub for technological innovation, business value creation, and talent development, contributing to the high-quality development of the digital economy.