"AI Low-Power Network to Reduce LAN Energy by 30% by 2027"

Mobile Korea 2024 Opens on the 30th
Baek Jeong-sook, ETRI Director: "Energy Reduction Through AI"
Reviewing Utilization of Domestic AI NPU Next Year

The Electronics and Telecommunications Research Institute (ETRI) is researching ways to reduce energy consumption in wireless access networks (LAN·RAN) by about 30% by 2027 through AI-based low-power network technology.

At 'Mobile Korea 2024' held on the 30th at LG Science Park in Gangseo-gu, Seoul, Bae Jeong-sook, head of ETRI's Intelligent Wireless Access Research Lab, stated, "ETRI is researching AI-based low-power network technology," adding, "Our goal is to reduce LAN energy consumption by about 30% by 2027."

Bae explained, "Reducing LAN usage is important because LAN accounts for 70% of the total network energy," and added, "We aim to reduce energy consumption through AI-based research and investigation."

The key is to classify the types of energy according to the amount of traffic being processed. This allows for analysis of which energy components need improvement. Bae said, "By distinguishing energy types, we can understand how much energy is consumed by each resource," and added, "With the development of AI network reduction technology, this technology can be applied in various traffic situations."

To increase computing power while using low-power networks, the introduction of AI semiconductors is also planned for next year. Bae stated, "We are working to collaborate with domestic neural processing unit (NPU) companies."