China Mobile and ZTE Unveil Industry-First Network Graph Model to Power Autonomous Networks

• China Mobile and ZTE launch the industry-first Network Graph Model, boosting root cause analysis accuracy to over 90% by integrating knowledge graphs with LLMs
• Powered by automated graph construction and multi-agent collaboration, the technology cuts MTTR by 20% and scales AI-driven O&M to 500,000 base stations by 2026
• The breakthrough enables Level-4 autonomous networks, transforming telecom operations from experience-based to data-driven intelligence

Copenhagen, Denmark, 25 June, 2026 - ZTE Corporation (0763.HK / 000063.SZ), a global leading provider of integrated information and communication technology solutions, today joined forces with China Mobile to unveil a groundbreaking Network Graph Model at DTW Ignite 2026. Developed through the China Mobile "Co-Innovation+" Autonomous Network Open Lab, the model integrates knowledge graphs with large language models (LLMs) to eliminate LLM hallucinations and deliver highly accurate, reliable, and scalable intelligent network analysis.

This industry-first innovation significantly enhances root cause analysis in multi-vendor, cross-domain networks. By shifting operations and maintenance (O&M) from experience-driven to data-driven, the technology marks a pivotal step toward high-level autonomous networks.

Network Graph Model Powers over 90% Root Cause Analysis Accuracy

To tackle persistent industry challenges such as heavy reliance on human expertise and difficulties in cross-domain fault diagnosis, China Mobile and ZTE have co-developed the Network Graph Model. The breakthrough bridges the gap between unstructured operational knowledge and massive volumes of structured alarm data, significantly enhancing the LLM's contextual understanding and logical reasoning in complex networks.

At its core is a real-time dynamic hybrid retrieval mechanism, capable of second search performance across knowledge graphs with tens of millions of nodes. For each fault event, the system dynamically constructs a scenario-specific reasoning sub-graph, which is then processed through deep LLM inference to trace causal relationships. In typical wireless macro and indoor distribution site outage scenarios, this approach has boosted root cause analysis accuracy from 50% to over 90%.

End-to-End Automated Graph Construction at Scale

To address the inefficiencies of traditional, manual knowledge graph construction, the joint team has pioneered a fully automated, end-to-end graph building process. This breakthrough reduces the time required to build knowledge graphs for 30,000 base stations from four days to just one, enabling rapid, large-scale deployment across new provinces.

The technology has been validated across more than 310,000 base stations in five provinces, including Beijing, Guangdong, and Shandong, demonstrating consistent root cause analysis accuracy exceeding 90%. Field data from Beijing reveals a 20% MTTR reduction and a 5% decrease in on-site technician visits, directly lowering O&M costs and enhancing user experience.

Multi-Agent Collaboration for Secure, Intelligent Networks

To further advance network autonomy, the team has introduced the Graph-Augmented A2A-T governance framework. This architecture strengthens multi-agent systems by enabling precise agent registration, discovery and full lifecycle governance. Through a trusted interaction architecture, intelligent agents can collaborate efficiently in complex networks, significantly improving task coordination success rates, operational efficiency, and system security. It lays a foundational architecture for the distributed autonomous network architectures envisioned for future 6G.

Scaling Joint Innovation to 500,000 Base Stations by 2026

The China Mobile "Co-Innovation+" Autonomous Network Open Lab serves as a pivotal hub for industry-academia-research-application integration. The collaboration with ZTE represents not only a core technological breakthrough in AI-driven networks but also delivers a replicable, scalable blueprint for intelligent operations transformation.

The newly launched High-Precision Network Graph Model marks a paradigm shift in AI's role within telecommunications, from AI-assisted to autonomous decision-making. The technology is set to expand to 10 provinces and 12 cities across China by 2026, covering over 500,000 base stations.

It will play a central role in enabling China Mobile to achieve Level-4 autonomous networks and offers a scalable model for global operators advancing digital transformation.