Flexible Reinforcement Learning Scheduler for 5G Networks

Resource management in 5G networks is becoming a highly complex task with the need to handle multiple devices and applications effectively with large data rates. In particular, optimizing scheduling algorithms improves resource allocation efficiency. To address this optimization, Artificial Intelligence (AI) techniques have recently shown their effectiveness compared to classical optimization algorithms. This work aims to design and implement a Reinforcement Learning (RL) MAC-layer scheduler that outperforms traditional methods in a potential real-world deployment. The main outcomes include a high-performing RL agent capable of adapting to diverse device types and handling a flexible number of user equipment (UE), which adds significant research value to the field since the current studies do not include a variable number of UEs.