Deep Reinforcement Learning-based UAV Control for Optimized Energy Efficiency and Throughput in UAV-assisted Communication

Jointly optimizing throughput and energy in unmanned aerial vehicle (UAV)-assisted networks remains a huge challenge at present. In this work, this problem is addressed with the powerful artificial intelligence method deep reinforcement learning (DRL). A communication model is firstly proposed which incorporates the advanced non-orthogonal multiple access communication technique for dealing with the intra-cluster interference in the network. Then, the optimization problem for jointly optimizing the throughput and energy of the UAV-assisted network considering quality of service (QoS) and other environmental constraints is formulated. Next, a Shared Dueling Deep Q Network (SDuDQN)-based DRL approach is proposed to solve the optimization problem. The simulation results show that the proposed approach can enhance both the throughput and energy efficiency as well as ensuring the spatial and QoS constraints for UAV-assisted networks.