The 5G bearer network, which connects the 5G radio access network and the core network and supports high quality private line services, plays an extremely important role. But troubleshooting on the bearer network can be difficult. On one hand, alarms and faults frequently occur. For example, a broken optical cable broken may trigger hundreds of device alarms. On the other hand, it typically takes several hours for experts to complete a fault diagnosis and the on-site engineer often needs to contact the network operations center to obtain support. Yet during typhoons and other disasters, emergency relief and communication recovery must be completed quickly.
This Catalyst is creating an intelligent fault management framework, encompassing network devices, the network management system (NMS) and the operations support system (OSS). The framework employs AI agents to automate the monitoring and diagnosis of root alarms, in place of manual operations, in common fault scenarios. In a scenario where a fault needs to be manually diagnosed, an AI copilot will provide support to the engineers via a natural language interface.
A major step towards the development of a level four autonomous network, the end-to-end solution is based on a three-layer architecture that associates digital twins with AI foundation models. Drawing on embedded AI, the intelligent network element (NE) layer provides real-time awareness of the network status. The intelligent NMS layer enables self-closed-loop fault diagnosis in a single domain. Integrated with the NMS, the intelligent OSS layer can address fault scenarios across domains and vendors end-to-end.
Having completed technical pre-research, the solution is being piloted by China Mobile Guangdong. After it is integrated into production, operations and maintenance in the province, the solution should greatly improve network stability and reliability, by reducing the time it takes to resolve faults. Improved data query efficiency and a more robust emergency response capability for natural disasters are also expected.