无人机辅助无蜂窝大规模MIMO中的空地协同调度  

Coordinated air-ground scheduling for UAV-assisted cell-free massive MIMO

作　　者：邓丹昊 王朝炜[1] 江帆 王卫东[1] DENG Danhao;WANG Chaowei;JIANG Fan;WANG Weidong(School of Electronic Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,China;School of Communication and Information Engineering,Xi’an University of Posts and Telecommunications,Xi’an 710121,China)

机构地区：[1]北京邮电大学电子工程学院,北京100876 [2]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《电信科学》2022年第8期37-44,共8页Telecommunications Science

基　　金：国家重点研发计划项目(No.2020YFB1807204)。

摘　　要：无蜂窝大规模多输入多输出(multiple-input multiple-output,MIMO)技术采用大量接入点(access point,AP)为地面用户提供高效的通信服务,但在有高速移动用户的场景中,会加剧对信道状态信息的依赖。为了减少导频资源的消耗,提出了一种综合无人机(unmanned aerial vehicle,UAV)辅助通信和无蜂窝大规模MIMO通信的双系统架构,该架构能够预测高速移动用户的轨迹,利用无人机为其提供可靠通信;进一步提出了基于深度强化学习(deep reinforcement learning,DRL)的无人机轨迹设计和地面用户调度方案,在满足各类约束的前提下实现系统总和速率最大化。仿真结果表明,与现有方案相比,所提方案能够有效提升系统容量。Cell-free massive multiple-input multiple-output(MIMO)technology uses a large number of access points(AP)to provide efficient communication services for terrestrial users.However,massive APs and users increase the demand for channel state information detection,especially for users with high moving speeds.In order to reduce the consumption of pilot resources,a dual system architecture that integrates unmanned aerial vehicle-assisted(UAV-assisted)communications and cell-free massive MIMO communications was proposed.The architecture was able to predict the movement trajectories of high-speed users and use UAVs to provide them with reliable communi-cations.A UAV trajectory design and user scheduling scheme based on deep reinforcement learning(DRL)was fur-ther proposed,which maximized the system sum rate under the premise of satisfying various constraints.Simulation results demonstrate that the proposed scheme is able to predict user trajectories and improve system sum rate com-pared with existing schemes.

关 键 词：无蜂窝大规模MIMO 无人机辅助通信 联合空地调度 深度强化学习 

分 类 号：TN915[电子电信—通信与信息系统]