A reinforcement learning approach to power control and rate adaptation in cellular networks

Optimizing radio transmission power and user data rates in wireless systems requires full system observability. While the problem has been extensively studied in the literature, practical solutions approaching optimality exploiting only the partial observability available in real systems are still lacking. This paper proposes a reinforcement learning approach to downlink power control and rate adaptation in cellular networks that closes this gap. We present a comprehensive design of the learning framework that includes the characterization of the system state, a general reward function, and an efficient learning algorithm. System level simulations show that our design quickly learns a power control policy that brings significant energy savings and fairness across users in the system.