Deep Reinforcement Learning Approach for Fairness-aware Scheduling in Wireless Networks

This paper studies a scheduling problem in wireless network systems which have combinatorial and time-varying properties. Although studies on efficient scheduling algorithms have been widely investigated, a huge trade-off between the performance and complexity still exists. Furthermore, it is becoming necessary to consider quality of service (QoS) and fairness constraints as growing demands on wireless networks. To this end, this paper propose a deep reinforcement learning (DRL) approach which can maximize system throughput while considering QoS and fairness by designing DRL structure according to the objective function and constraints. Numerical results demonstrate the effectiveness of the proposed approach.