Scholarlink

<p>With the growing demand for large-scale Internet of Things (IoT) connectivity, unmanned aerial vehicle (UAV) base stations offer flexible means to enhance network capacity. This paper investigates the joint optimization of the deployment of multiple UAVs and the uplink access of multiple user equipment (UEs) in non-orthogonal multiple access (NOMA) networks. We formulate system energy efficiency as the primary objective and propose a hierarchical reinforcement learning framework to solve the resulting multi-timescale optimization problem. The framework decomposes decisions into two levels, with a high-level policy that determines the number and placement of UAVs on a slower timescale and a low-level policy that, on a faster timescale, coordinates UE access selection and resource allocation. A hierarchical variant of Proximal Policy Optimization (PPO) is adopted to enable coordinated end-to-end training of both policy levels and to handle the coupling between deployment and access-control decisions. Simulation results demonstrate that the proposed scheme enables end-to-end joint optimization of deployment and access control, and achieves higher energy efficiency and throughput than conventional two-stage baselines. Further analysis validates its effectiveness for large-scale IoT uplink access scenarios.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Joint Optimization of UAV Deployment and Resource Allocation for Energy Efficiency in NOMA-Enabled UAV Networks

  • 期刊论文

  • 发表日期：

    2026年4月27日

查看全文
  • Xiang Wei,
  • Jie Ding,
  • Ziqing Weng,
  • Xu Zhao

摘要

With the growing demand for large-scale Internet of Things (IoT) connectivity, unmanned aerial vehicle (UAV) base stations offer flexible means to enhance network capacity. This paper investigates the joint optimization of the deployment of multiple UAVs and the uplink access of multiple user equipment (UEs) in non-orthogonal multiple access (NOMA) networks. We formulate system energy efficiency as the primary objective and propose a hierarchical reinforcement learning framework to solve the resulting multi-timescale optimization problem. The framework decomposes decisions into two levels, with a high-level policy that determines the number and placement of UAVs on a slower timescale and a low-level policy that, on a faster timescale, coordinates UE access selection and resource allocation. A hierarchical variant of Proximal Policy Optimization (PPO) is adopted to enable coordinated end-to-end training of both policy levels and to handle the coupling between deployment and access-control decisions. Simulation results demonstrate that the proposed scheme enables end-to-end joint optimization of deployment and access control, and achieves higher energy efficiency and throughput than conventional two-stage baselines. Further analysis validates its effectiveness for large-scale IoT uplink access scenarios.