2. What is AI for RAN in AI-RAN?
The AI-RAN proposed by SoftBank virtualizes sufficient computing processing power of hardware resources within data centers, on which RAN and AI applications are superimposed. By achieving flexible allocation of resources in response to traffic, etc., it aims to optimize equipment investment and operational efficiency.
Moreover, by consolidating a large number of base stations and layering RAN and RIC/AI on the same platform, it becomes possible to solve problems such as data transfer tasks and insufficient computing capacity.
The overlay of RAN and RIC/AI enables real-time RAN control in the millisecond to microsecond range, a feat difficult to achieve even with Near-RT-RIC. Previously, the allocation of MCS and resource blocks was carried out based on a specific threshold or algorithm derived from the uplink radio signal information coming from the terminal. By integrating AI-based prediction and real-time data analysis, it is possible to understand the wireless environment in real-time and make optimal MCS and resource block assignments. This optimizes wireless scheduling and therefore improvements in user throughput and user experience are expected.
By collecting data from many base stations and implementing learning and control via AI/ML, coordination between multiple base stations can be achieved. This allows area-wide control of base stations, such as adjusting transmission power, beamforming, suppressing interference waves, etc. This also enables removal of packet congestion, improvement of MIMO rate, realization of flexible CA combinations, leading to improved throughput and user experience. .
Not only can it use computing resource utilization and information from the RAN, but by coordinating with external data such as time, weather, event information and performing statistics or learning, it can create what is called a digital twin. This makes it possible to predict user behavior via AI/ML and pre-realize optimal RAN settings.
