Scholarlink

This paper investigates the computation offloading and resource scheduling problem for STAR-RIS-enabled NOMA-ISCC systems, aiming to minimize long-term system energy consumption under stringent latency constraints. The study first establishes a dynamic system energy consumption model. To effectively address this complex optimization problem, a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm framework is proposed, where multi-agent task decomposition significantly enhances policy exploration efficiency and convergence speed. The simulation results demonstrate that the proposed scheme exhibits substantial energy efficiency advantages under various conditions, achieving significant reductions in the energy consumption of the system (average reductions of 18.37%, 52.11%, 62.86% and 12.45%, 33.71%, 50.62%, respectively). These findings validate the superiority of the MADDPG algorithm compared to benchmark schemes and confirm its effectiveness in optimizing long-term system energy consumption.

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

Resource Scheduling in Dynamic STAR-RIS-Enabled NOMA-ISCC Systems Based on Multi-agent Deep Reinforcement Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shangxiao Wu,
  • Yiting Huang,
  • Zhaohui Chen

摘要

This paper investigates the computation offloading and resource scheduling problem for STAR-RIS-enabled NOMA-ISCC systems, aiming to minimize long-term system energy consumption under stringent latency constraints. The study first establishes a dynamic system energy consumption model. To effectively address this complex optimization problem, a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm framework is proposed, where multi-agent task decomposition significantly enhances policy exploration efficiency and convergence speed. The simulation results demonstrate that the proposed scheme exhibits substantial energy efficiency advantages under various conditions, achieving significant reductions in the energy consumption of the system (average reductions of 18.37%, 52.11%, 62.86% and 12.45%, 33.71%, 50.62%, respectively). These findings validate the superiority of the MADDPG algorithm compared to benchmark schemes and confirm its effectiveness in optimizing long-term system energy consumption.