ETRI Unveils Natural-Language Optical Network Configuration and Pre-Deployment Verification

According to AI TIMES, ETRI on October 14 unveiled an AI operations technology that automatically handles optical access network equipment registration and configuration using only a user's natural-language commands, while also checking for possible errors in a virtual environment before actual deployment. The research team also presented the results of its technology accumulation by announcing a total of 5 papers at related international conferences over the past 2 years. The new technology focuses on reducing the need for people to manually learn and enter device-specific commands and procedures, instead having AI interpret intent, organize the sequence of tasks, and extend to device control. In particular, it makes it possible to precheck control procedures in an environment that reflects the current network state before execution, laying the groundwork for filtering out incorrect settings before they are applied to the actual network. Another feature is the development of a core local LLM element specialized for optical access networks that can run in-house instead of relying on external commercial models, helping reduce concerns that operational information could be sent outside. Based on this, ETRI plans to advance optical access network automation beyond command execution toward a direction in which the network itself understands and judges its own state.