No abstract

No abstract

Unmanned aerial vehicle (UAV)-assisted communications is a promising solution to improve the performance of future wireless networks, where UAVs are deployed as base stations for enhancing the quality of service (QoS) provided to ground users when traditional terrestrial base stations are unavailable or not sufficient. An effective framework is proposed in this paper to manage the dynamic movement of multiple unmanned aerial vehicles (UAVs) in response to ground user mobility, with the objective to maximize the sum data rate of the ground users. First, we discuss the relationship between the air-to-ground (A2G) path loss (PL) and the location of UAVs. Then a deep Q-network (DQN) based method is proposed to adjust the locations of UAVs to maximize the sum data rate of the user equipment (UE). Finally, simulation results show that the proposed method is capable of adjusting UAV locations in a real-time condition to improve the QoS of the entire network.Index Terms-Unmanned aerial vehicle (UAV), UAV-assisted network, reinforcement learning, user equipment (UE), quality of service (QoS). I. INTRODUCTIONT HE unprecedented demand for high-quality wireless communications has fueled the evolution of wireless technologies and communications networks. The unmanned aerial vehicle (UAV)-assisted network where UAVs are deployed and function as aerial base stations to assist the terrestrial base stations is an effective complementary solution to emergency wireless service recovery after natural disasters or infrastructure damage [1]. Also, in Internet of Things (IoT) networks, UAVs can be used as aerial base stations to collect data from ground devices, in which building a complete cellular infrastructure is not affordable [2]. The authors in [3] proposed a multi-layer UAV network model for UAV-enabled 5G and beyond applications. Despite advantages such as flexibility, mobility, cost and time efficiency in UAVassisted networks, one key design challenge is to determine the move strategy for UAVs. Since in realistic situations, the environment where UAVs are deployed is highly dynamic, it is critical for UAVs to adjust its locations regularly to cope with varying conditions. Furthermore, utilizing machine learning techniques for the UAV communication recently has seen unprecedented growing popularity [4].