Scholarlink

Satellite-aerial-terrestrial integrated networks (SATINs) with reconfigurable intelligent surface (RIS) and non-orthogonal multiple access (NOMA) are promising for next-generation communications due to their ability to intelligently reconfigure transmission environments and improve spectral efficiency, but face the tricky challenge of multi-objective optimization of network performance in dynamic communication environments. To better describe the model, a multi-vector optimization problem (MVO) is proposed to improve system reachability under many constraints, considering UAV energy, transmit beamforming, and RIS phase-shift. Real-time interaction with the environment is achieved with DRL algorithms, using multi-vector deep deterministic policy gradient (MV-DDPG) to find a suboptimal solution. Experimental results show an effective adjustment of the optimal update policy when considering three objectives simultaneously and different weighting settings.

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

NOMA-Enabled RIS-Assisted SATINs Under Multi-objective Optimization: A Deep Reinforcement Learning Approach

  • 章节

  • 发表日期：

    2025年

查看全文
  • Kefeng Guo,
  • Min Wu,
  • Xingwang Li,
  • Ali Nauman,
  • Changqing Li,
  • Mingfu Zhu

摘要

Satellite-aerial-terrestrial integrated networks (SATINs) with reconfigurable intelligent surface (RIS) and non-orthogonal multiple access (NOMA) are promising for next-generation communications due to their ability to intelligently reconfigure transmission environments and improve spectral efficiency, but face the tricky challenge of multi-objective optimization of network performance in dynamic communication environments. To better describe the model, a multi-vector optimization problem (MVO) is proposed to improve system reachability under many constraints, considering UAV energy, transmit beamforming, and RIS phase-shift. Real-time interaction with the environment is achieved with DRL algorithms, using multi-vector deep deterministic policy gradient (MV-DDPG) to find a suboptimal solution. Experimental results show an effective adjustment of the optimal update policy when considering three objectives simultaneously and different weighting settings.