ETRI Leads Research on Machine Learning-Based Future Frequency Standards

ETRI researchers discussing a machine learning-based spectrum availability prediction method. (Clockwise from top left) Researcher Jonghoon Yoon, Director Seunggeun Park, Researcher Yunbae Kim, Researcher Hyeyeon Kwon.

[Asia Economy Reporter Eunmo Koo] A machine learning-based method developed by domestic researchers for analyzing and predicting frequency usage is expected to become an international standard.

The Electronics and Telecommunications Research Institute (ETRI) announced on the 30th that at the ‘International Telecommunication Union Radiocommunication Sector Spectrum Management Study Group (ITU-R SG1) meeting,’ held over two weeks starting from the 25th of last month, ETRI’s ‘Machine Learning-Based Spectrum Availability Prediction Method’ was adopted as a draft working document for a new report.

'Spectrum availability' is an indicator representing the extent to which radiocommunication services can be used within a specific frequency band. Accurate analysis of this is essential to understand frequency usage and saturation levels. Unused frequencies can be reclaimed and reassigned, enabling efficient management of radio resources.

Until now, analysis was conducted using simplified mathematical models, making it difficult to analyze complex radio environments. Additionally, there was no standard document organizing analysis methods by various frequency types and usage patterns, leaving no guidelines to refer to.

ETRI systematically organized spectrum availability analysis methods by frequency type and usage behavior. Furthermore, focusing on the most commonly used mobile communication frequencies, they proposed a methodology that reflects complex radio environments and evaluates and predicts spectrum availability using machine learning.

The research team evaluated and predicted the availability of LTE frequency spectrum in South Korea using this method. Based on measured traffic data, they assessed the supply-demand balance of LTE frequencies and predicted future usage rates, marking the world’s first result of its kind, which is highly significant.

The research team had already proposed this working document-related research task at the 2019 ITU-R SG1 meeting and, after undergoing the UN member state circulation process, received approval as a new research task for SG1.

Seunggeun Park, Head of the Radio Resource Research Division at ETRI, said, “We will continue to strive to secure leadership in next-generation mobile communication frequency research based on the data and machine learning-based spectrum availability evaluation and prediction methods developed this time.”

Going forward, the research team plans to actively participate in international standardization efforts related to radio and information communication fields, while continuously researching analysis methodologies suited to South Korea’s radio environment to support the establishment of effective national radio resource management policies.

This content was produced with the assistance of AI translation services.