Scholarlink

<p>With the rapid advancement of emerging technologies such as the metaverse and digital twins, efficient coordination among communication, computation, and storage in edge computing environments has become increasingly important. This research presents a novel industrial metaverse architecture based on digital twins, which optimizes resource utilization by leveraging mobile edge computing and ultra-reliable low-latency communications. The proposed framework employs task offloading and edge storage to reduce latency and satisfy the stringent requirements of future metaverse systems in terms of reliability and delay minimization. The proposed method utilizes reinforcement learning algorithms, including Deep Q-Network (DQN) and its advanced variants, namely Double Deep Q-Network (DDQN) and Dueling Deep Q-Network (Dueling DQN), to enable intelligent decision-making and adaptability in dynamic environments. Simulation results demonstrate that the proposed approach consistently outperforms baseline methods, reducing end-to-end latency by 10–15% compared to state-of-the-art techniques.</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

An edge intelligence framework with reinforcement learning for digital twins in industrial metaverse

  • 期刊论文

  • 发表日期：

    2026年7月23日

查看全文
  • Reza Mohammadvand,
  • Nasser Mozayani,
  • Saeed Khoshkholghi

摘要

With the rapid advancement of emerging technologies such as the metaverse and digital twins, efficient coordination among communication, computation, and storage in edge computing environments has become increasingly important. This research presents a novel industrial metaverse architecture based on digital twins, which optimizes resource utilization by leveraging mobile edge computing and ultra-reliable low-latency communications. The proposed framework employs task offloading and edge storage to reduce latency and satisfy the stringent requirements of future metaverse systems in terms of reliability and delay minimization. The proposed method utilizes reinforcement learning algorithms, including Deep Q-Network (DQN) and its advanced variants, namely Double Deep Q-Network (DDQN) and Dueling Deep Q-Network (Dueling DQN), to enable intelligent decision-making and adaptability in dynamic environments. Simulation results demonstrate that the proposed approach consistently outperforms baseline methods, reducing end-to-end latency by 10–15% compared to state-of-the-art techniques.