Reinforcement Learning-Based Radio Resource Control in 5G Vehicular Network

Recently, the number of user equipment with high mobility (such as vehicles) and the high traffic demand is immensely increasing. To sustaining the traffic demand, Time Division Duplex (TDD), which can change the uplink and downlink ratio in the same frequency band, is implemented. However, to satisfy the high traffic demand in 5G vehicular network with limited resources, we need to investigate a more efficient method to adaptively change TDD configuration. In this letter, we propose a reinforcement learning-based resources allocation algorithm that not only considers the multiple parameters of network status but also utilizes them in the learning phase. The contribution of this letter is that our proposal takes into account the future state for resource control through the learning phase. The simulation results show that our proposal outperforms the existing method in throughput and packet loss rate.