Reinforcement Learning Framework for Dynamic Power Transmission in Cloud RAN Systems

5G and beyond communication systems suffer from high path loss which leads to low coverage area. Weather elements such as rain, fog, and clouds aggravate the issue of low coverage area, and in addition, cause a reduction in throughput. We develop a reinforcement learning (RL) based solution that enables coordination among the radio access network (RAN) systems for dynamically adjusting power transmission. Our solution reduces the interference and increases the throughput of the system while simultaneously increasing the coverage area. We achieve a 28% gain in throughput across multiple base stations under rainy conditions. The use of our RL enabled near-optimal framework offers strong benefits in 5G, mmWave and cmWave systems, and in addition, offers the possibility of incorporation in any state-of-the-art technology such as cloud, virtual, or open RAN systems (CRAN, VRAN or O-RAN).