Scholarlink

To meet the low-latency routing requirements of a large number of emerging applications in ultra-reliable communication environments, this paper introduces a deep reinforcement learning (DRL)-based low-latency high-reliability intelligent routing scheme for software-defined networking (SDN). We formulate a minimum decision latency model to describe the optimal routing problem based on a comprehensive overview of network resources and centralized control of network devices from SDN, where dynamic differentiated routing requirements are considered. A DRL-based routing algorithm is presented to achieve ultra-reliable low-latency routes by integrating the ultra-reliable routing requirement module responsible for the unqualified routes filter. Experiment results show that the proposed algorithm outperforms in routing reliable requirements and decision latency compared to other popular algorithms, especially in large-scale networks and high-dynamic network scenarios.

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

Deep Reinforcement Learning-Based Decision Latency Optimization for Ultra-Reliable Communication Routing

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Jiabao Chen,
  • Mukun Chen,
  • Xige Zhang,
  • Dongyang Wang,
  • Jinyi Chen,
  • Zhaoming Hu,
  • Zhuwei Wang,
  • Chao Fang

摘要

To meet the low-latency routing requirements of a large number of emerging applications in ultra-reliable communication environments, this paper introduces a deep reinforcement learning (DRL)-based low-latency high-reliability intelligent routing scheme for software-defined networking (SDN). We formulate a minimum decision latency model to describe the optimal routing problem based on a comprehensive overview of network resources and centralized control of network devices from SDN, where dynamic differentiated routing requirements are considered. A DRL-based routing algorithm is presented to achieve ultra-reliable low-latency routes by integrating the ultra-reliable routing requirement module responsible for the unqualified routes filter. Experiment results show that the proposed algorithm outperforms in routing reliable requirements and decision latency compared to other popular algorithms, especially in large-scale networks and high-dynamic network scenarios.