Reinforcement Learning for Radio Resource Management in O-RAN

Open RAN is a groundbreaking approach that trends to revolutionize mobile network communication. Open RAN steers in a new era of data-driven decision-making RAN management by introducing intelligent controllers that adapt the Radio Access Network (RAN) in near-real time. As a consequence of the current surge in network traffic, the role of Artificial Intelligence in optimizing the RAN becomes paramount. The fifth-generation mobile networks (5G) further enhance this potential by enabling the allocation of clients through network slices for services such as eMBB, URLLC, and mMTC services. This paper presents a radio resource allocation policy that leverages Reinforcement Learning to make intelligent decisions that align with the objectives of each service in the network. The proposed innovative approach reduces user allocation waiting time by 20% compared to the traditional Round-Robin (RR) policy.