Intelligent reinforcement learning-based scheduling in 5G networks and beyond

The emergence of fifth-generation (5G) wireless networks has brought forth notable enhancements in cellular technology, supporting rapid data transfer rates, reduced latency, and broad connectivity. Yet, the efficient allocation of resources continues to be a significant hurdle, fueled by the varied demands for quality of service (QoS), ever-changing channel dynamics, and the necessity for fairness among users. As a solution, reinforcement learning (RL) has risen as a potent method for adept scheduling within these networks, capitalizing on its proficiency to deduce the best strategies without reliance on predefined mathematical frameworks and its flexibility in adapting to shifting circumstances. This chapter explores the application of RL for scheduling in 5G and beyond, focusing on its potential to solve resource allocation complexities. It covers basic RL strategies such as Markov decision processes (MDPs), Q-learning, and deep Q-networks (DQN), and details their application in developing scheduling solutions. The text highlights RL’s significant advancements over traditional methods through case studies and discusses challenges and opportunities in Beyond 5G (B5G) and 6G networks. It also examines emerging technologies like the Internet of Things (IoT) and unmanned aerial vehicles (UAVs), and introduces advanced techniques like transfer learning and multi-agent RL to address scalability and complexity. Finally, it outlines key research areas, including adaptive management in dynamic settings and the integration of various artificial intelligence (AI) and machine learning (ML) techniques for enhanced scheduling solutions.