AI Sets Up Optical Networks on Command
TECHWORLD ·
✦ AI Summary
ETRI announced the development of "LLM-based optical access network operation technology."
When a user enters an operational goal in natural language, AI analyzes the intent of the command, plans the procedures, and controls the optical access equipment.
ETRI is also developing preverification technology that checks systems in a virtual environment before actual deployment, as well as core local LLM technologies that do not rely on external commercial AI.
The Electronics and Telecommunications Research Institute (ETRI) announced the development of "LLM-based optical access network operation technology." The technology enables an LLM to understand a user's natural-language instructions. That means operators no longer need to enter complex commands for each device or input operational procedures one by one.
With this technology, the LLM automatically configures the optical access network. It also allows errors to be checked in advance before actual deployment. As a result, the system makes it possible to operate an optical access network using natural language alone.
ETRI said the technology is expected to serve as a foundation for building an "AI-native" optical access network. An AI-native optical access network refers to a network management method based on AI's awareness of network conditions and decision-making. ETRI said the new technology is expected to be used to implement that approach.
An optical access network is a network that delivers ultra-high-speed communication services to user-adjacent points such as homes, businesses, and mobile base stations. It is commonly referred to as the "capillaries of the communications network."
The technology developed by ETRI works by having the user enter network operation goals in natural language, after which AI analyzes the intent of the command, autonomously plans the required work procedures, and then controls the actual optical access equipment. Its application targets real network operation procedures such as equipment registration and configuration.
ETRI conducted interoperability tests with optical access equipment. The results confirmed that network operation could be possible using natural language alone.
ETRI also developed a "preverification technology." This technology checks AI-generated control procedures in a virtual environment before they are applied to the actual optical access network.
ETRI is also paying attention to security issues arising from the use of commercial LLMs. That is because if an external commercial LLM is used for optical access network operations, key network operation information and data could be transmitted outside the network. Accordingly, ETRI is developing core technologies for implementing a "local LLM specialized for optical access" that is intended to run independently inside the network without relying on external commercial AI.
The research team said it has secured key AI technologies needed for operating optical access networks. The technologies secured include transferring the knowledge and capabilities of large-scale AI models to smaller local LLMs, improving the efficiency of adapting local LLMs to specific optical access network operations, and evaluating whether AI-generated results are accurate and meet user requirements.
Jo Seung-hyun, head of ETRI's wired and wireless optical access research lab, said the research has laid the technical foundation for AI to understand operators' natural-language commands, carry out required tasks, and verify results before actual deployment. He added that the institute plans to pursue the advancement of AI-native optical access technology centered on a future "optical access foundation LLM," with the goal of enabling AI to autonomously perceive, judge, and control network conditions.
Source: TECHWORLD · Lee Gwang-jae
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408161
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.