SoftBank Corp. deploys AI system to optimize Massive MIMO

SoftBank Corp. and Ericsson Japan K.K. demonstrated and deployed an AI-powered, externally controlled optimization system for Massive Multiple-Input Multiple-Output (MIMO), and the implementation produced a 24% increase in 5G downlink throughput during periods of high traffic fluctuation.

Following trials at Expo 2025, the companies began rolling out the system at large event venues, including major arenas and dome-type facilities in the Tokyo metropolitan area, with the stated purpose of managing heavily fluctuating traffic patterns.

The deployed solution used an external control server that collected user distribution and Massive MIMO beam estimation data from base stations at one-minute intervals; an Artificial Intelligence (AI) model had been trained using performance results from multiple prechanged coverage patterns and automatically determined event occurrence to adjust horizontal and vertical beam patterns every minute. The system also reacted to sudden traffic surges to mitigate what the release described as “packet stalling.”

During demonstrations, the controlled coverage adjustments switched base station patterns to configurations identified by the AI and increased downlink speeds from 76.9 Mbps to 95.5 Mbps under high traffic fluctuation, representing a 24% improvement over the measured baseline.

SoftBank Corp. is expanding into non-telecom fields under its “Beyond Carrier” growth strategy while growing its telecom business with 5G/6G, Internet of Things (IoT), Digital Twin, and Non-Terrestrial Network solutions including High Altitude Platform Station (HAPS), and it is constructing AI data centers, developing homegrown LLMs for the Japanese language, and integrating AI with radio access networks (AI-RAN) with the aim of becoming a provider of next-generation social infrastructure.