AI-RAN technology is moving closer to real-world mobile networks as Samsung Electronics and NTT DOCOMO test an AI-driven system designed to predict service degradation and automatically optimize network settings for individual users.
The two companies said they successfully verified a user-specific network optimization technology that combines artificial intelligence with the radio access network, or RAN. The work represents another step toward AI-native mobile infrastructure, where networks could increasingly adapt to changing conditions without relying entirely on static configurations.
The technology analyzes factors including a user’s movement, service usage patterns and real-time radio conditions. AI models then attempt to identify signs of declining service quality before problems become noticeable and select network settings intended to maintain connectivity and performance. citeturn1view0
AI-RAN Technology Targets Individual Network Conditions
Traditional mobile networks generally apply common network configurations to smartphones connected to the same base station.
That approach can become less effective when individual users experience significantly different radio conditions. Someone moving through an area with weaker coverage, for example, may experience reduced speeds or interrupted connectivity even though other devices connected to the same base station perform normally.
Samsung and NTT DOCOMO’s approach is designed to make those decisions more individualized.
The AI-RAN technology learns patterns associated with individual users and attempts to predict when service quality is likely to deteriorate. It can then select network parameters, including an appropriate frequency band, before the expected degradation occurs. citeturn1view0
For video streaming, this could mean detecting conditions likely to cause slower transmission speeds, buffering or reduced image quality and adjusting network settings before those effects become significant.
Tests Show Lower Rate of Service Degradation
Samsung and NTT DOCOMO have been working together across multiple stages of the project, including AI-RAN development, network data collection and improvements to data measurement procedures.
The companies conducted simulation testing using data collected in January from NTT DOCOMO’s commercial network and a local 5G test network in Japan.
According to Samsung, the tests reduced the frequency of service throughput degradation from 13.1% to 7.2% — roughly cutting the occurrence rate in half. citeturn1view0
The result is significant because AI-RAN is increasingly being explored not simply as a way to automate network operations, but as a technology that could directly influence the experience of individual mobile users.
The test results were reported by Samsung and should be viewed as a technical validation rather than evidence of performance across every commercial network environment.
AI Could Make Mobile Networks More Adaptive
Radio access networks are a critical part of mobile infrastructure because they connect devices such as smartphones to the wider telecommunications network.
Samsung has been researching ways to integrate AI across different parts of the RAN. The company previously demonstrated AI-RAN technologies aimed at improving network performance and energy efficiency, describing AI integration as an important part of the evolution toward future communications. citeturn1search1
NTT DOCOMO is pursuing similar research as the telecommunications industry prepares for 6G.
In March 2026, DOCOMO and VIAVI reported a separate demonstration of AI-driven RAN control for future 6G networks, while DOCOMO and SK Telecom have also published research covering the evolution of virtualized RAN and the path toward AI-RAN. citeturn1search5turn1search0
Together, these projects illustrate a broader industry shift toward networks capable of using AI to analyze conditions and dynamically adjust their behavior.
Reducing the Data Burden of AI-RAN
Introducing more AI into mobile networks also creates another challenge: collecting and processing the data required by AI models can itself consume network and computing resources.
Samsung and NTT DOCOMO said they developed a data-processing method intended to reduce that burden.
Instead of collecting and processing all available radio-environment information, the approach selects data according to the specific problem a user is experiencing. The goal is to provide the AI system with relevant information while limiting unnecessary network overhead. citeturn1view0
The companies have also taken their work beyond laboratory development.
Samsung said the two partners jointly submitted the network data-processing procedure to 3GPP, the global mobile communications standards organization, in February as part of their ongoing cooperation on technologies relevant to future 6G networks. citeturn1view0
AI-RAN Could Become an Important Building Block for 6G
AI-RAN is emerging as one of several technologies being explored for the transition from today’s 5G infrastructure toward more intelligent future networks.
Rather than simply increasing peak transmission speeds, future mobile networks are expected to become more adaptive, using AI to optimize resources, improve efficiency and respond to individual service requirements.
Industry work remains at various stages of research, testing and standardization. DOCOMO and SK Telecom noted in their 2026 AI-RAN white paper that using virtualized base stations to provide AI computing capabilities remains at an early stage. citeturn1search22
That makes commercial-network testing particularly important.
Samsung and NTT DOCOMO’s latest results suggest that AI-RAN technology could eventually allow mobile networks to respond more precisely to the conditions experienced by individual users rather than relying primarily on uniform network configurations.
Further testing, standardization and commercial validation will be necessary before such capabilities become commonplace. But as telecom operators and equipment vendors prepare for 6G, AI-driven network optimization is increasingly becoming part of the industry’s roadmap.
