基于深度强化学习的空中基站部署优化算法  

Aerial Base Station Deployment Optimization Based on Deep Reinforcement Learning

作　　者：张博[1] 杨锟浩 李俊锋[1] ZHANG Bo;YANG Kunhao;LI Junfeng(School of Cyber Science and Engineering,Zhengzhou University,Zhengzhou 450002,China)

机构地区：[1]郑州大学网络空间安全学院,河南郑州450002

出　　处：《郑州大学学报（理学版）》2024年第4期28-33,共6页Journal of Zhengzhou University:Natural Science Edition

基　　金：国家自然科学基金面上项目(61972092);郑州市协同创新重大专项(20XTZX06013)。

摘　　要：针对无人机群(swarm of unmanned aerial vehicle,UAV-swarm)在救灾场景中对地面移动用户进行持续性通信覆盖的问题,设计了一种基于多智能体的深度强化学习的无人机群路径优化算法。该算法框架中无人机具有分布式决策能力,根据用户的移动来动态调整自身的移动策略。通过设置合适的强化学习奖励和参数,使无人机在满足覆盖百分比、防碰撞、能源限制等多种约束前提下,最大限度地长期覆盖地面移动用户。与其他无人机部署方案算法进行仿真对比,实验结果表明,该模型在收敛速度和收敛效果上得到了显著提升。To ensure of continuous communication coverage of ground mobile users by UAV-swarm clusters in disaster relief scenarios,a UAV-swarm cluster path optimization algorithm was designed based on deep reinforcement learning of multiple intelligences.The UAVs in this algorithm framework had distributed decision making capability,and could dynamically adjust their own movement strategy according to the user′s movement.The UAVs should be deployed to maximize long-term coverage of ground mobile users in a specified area by setting appropriate reinforcement learning rewards and parameters while satisfying multiple constraints such as coverage percentage,collision avoidance,and energy constraints.The proposed model was compared with other UAV deployment scheme algorithms by simulation.Results showed that the model significantly improved in terms of convergence speed and convergence effect.

关 键 词：无人机 无线通信 深度强化学习 轨迹规划 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]