Power Allocation based on Reinforcement Learning for QoS Support in MIMO systems

This paper proposes a novel reinforcement learning-based power allocation scheme, a unique approach to support QoS in MIMO systems. In MIMO systems, efficient power allocation is critical to mitigate inter-device interference and improve energy efficiency. Given the dynamic nature of the wireless channel environment and the potential changes in QoS levels, a dynamic channel allocation scheme that can satify the QoS with minimum power consumption is essential. Our proposed scheme, which allocates power to each antenna based on reinforcement learning, stands out for its ability to reduce unnecessary search range by varying the reward value according to the channel environment and QoS satisfaction estimated by pilot signals. Simulations further validate the feasibility and superiority of our innovative scheme.