As 5G adoption accelerates, transport networks are generating millions of alarms each week, placing significant pressure on telecom operations teams. Traditional, manual troubleshooting processes struggle to keep pace, often taking two to three hours to fully resolve faults and relying heavily on the experience of individual engineers.<\/p>\n\n\n\n
To address this challenge, China Mobile Zhejiang<\/strong>, working with ZTE<\/strong>, deployed an AI-driven automated troubleshooting solution that combines large language models (LLMs)<\/strong>, small AI models<\/strong>, and a digital twin of the transport network<\/strong>. The solution replaces rule-based alarm handling with intelligent correlation, root-cause analysis, and closed-loop fault resolution.<\/p>\n\n\n\n The system automatically filters and clusters alarms, identifies root causes with over 90% accuracy, and optimises the entire fault management workflow, from identification and diagnosis to repair. Configuration-related faults can be resolved fully automatically via simulations in the digital twin, while device-related faults are supported with real-time repair guidance delivered to field engineers through a mobile app, eliminating delays caused by manual coordination.<\/p>\n\n\n\n Results from the deployment:<\/strong><\/p>\n\n\n\n Beyond cost and efficiency gains, the solution improves network reliability and service quality for 4G\/5G and dedicated enterprise lines, supporting critical sectors such as government, finance, and manufacturing. ZTE positions the solution\u2019s integrated \u201calgorithm-platform-process\u201d design as a reusable framework that can be replicated across operators, network domains, and even other industries undergoing digital transformation. <\/p>\n\n\n\n \n