Presentation Information
[BI-3-04]Automation of Network Fault Analysis and Prospects for Autonomous Operations
〇Shogo Fukushima1, Sora Funayama1, Junta Tachibana1, Masako Takahashi1 (1. NTT DOCOMO, Inc.)
Keywords:
AIOps,Network Fault Analysis,Autonomous Networks,Large Language Models (LLMs),Retrieval-Augmented Generation (RAG)
With the increasing complexity of telecommunications networks, improving the efficiency and sophistication of fault management operations has become increasingly important. This paper proposes an AI agent that leverages Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs) to automate root cause analysis and corrective action guidance in commercial network operations. The proposed approach applies LLM-based preprocessing to operational knowledge accumulated through past maintenance activities, thereby constructing a knowledge base optimized for retrieval and reasoning. In addition, an orchestrator coordinates multiple tools to perform fault analysis and generate recommended corrective procedures. Evaluation using alerts generated in a commercial network environment demonstrated that the proposed AI agent can provide analysis results and corrective action procedures comparable to those produced by experienced network operators, while reducing analysis time and delivering consistent response performance. Furthermore, the proposed approach is examined in the context of the TM Forum Autonomous Networks framework, highlighting its potential contribution to the realization of autonomous network operations.