Samsung Electronics has increased download speeds by up to 52% on Japanese mobile operator KDDI's commercial 5G Standalone (SA) network using an artificial intelligence (AI)-based network optimization technology. With commercial 5G SA services also expected to expand in South Korea, attention is turning to whether the technology can significantly improve network performance.
Samsung Electronics and KDDI announced on July 1 that they had successfully demonstrated RAN Speed Optimizer (RSO), an AI-powered radio access network (RAN) optimization solution, in Tokyo and surrounding areas of Japan.
The trial was conducted over several months beginning late last year on KDDI's live commercial 5G SA network. The companies evaluated the technology across hundreds of network cells deployed in dense urban, suburban, and rural environments. The tests used 100 MHz of Time Division Duplex (TDD) spectrum in the 3.7 GHz band. By operating on a commercial network serving real subscribers, the AI model was trained and validated under varying traffic conditions across different locations and times of day.
The results showed that average 5G download throughput during peak traffic hours increased by 31% across all test areas. In densely populated urban environments with high concentrations of buildings and users, download speeds improved by as much as 52%.
RSO is an AI prediction model developed by Samsung that analyzes traffic patterns and the radio environment surrounding each base station to recommend optimal configuration settings for individual cells. Unlike conventional approaches that apply the same settings across multiple cells, RSO tailors configurations to each cell by considering factors such as building density, user distribution, and radio interference, enabling the network to adapt to changing traffic conditions by location and time.
The technology reduces the need for network engineers to manually adjust base station parameters. It also improves network capacity and efficiency without requiring additional hardware or new spectrum resources, allowing operators to make better use of existing infrastructure. AI- and machine learning-based RAN automation and performance optimization are widely regarded as key technologies for Open RAN and next-generation network operations.
RSO is part of Samsung's AI-driven network management platform, the Samsung Cognitive Network Operations Suite (NOS). The platform includes network automation solutions, AI agents, and intelligent applications designed to optimize network operations.
Building on the successful trial, Samsung Electronics and KDDI plan to expand testing of AI-based optimization technologies across commercial networks. The companies also expect the technology to be deployed commercially in other markets, including South Korea, where 5G SA services are being rolled out.
Moon Jun, Executive Vice President and Head of R&D at Samsung Electronics' Networks Business, said, “Since 2024, we have been testing the RSO technology in real-world environments and continuously training the AI model. This trial demonstrates that AI-based optimization can deliver stable connectivity and improved network performance across diverse commercial network conditions.”
