Scholarlink

<p>The growth of IoT applications increases the demand for scalable computing resources. While cloud computing provides high processing and storage capacity, its centralized nature can cause latency and network congestion. Fog computing mitigates these issues by bringing computation closer to IoT devices, but its limited resources require efficient task scheduling to optimize performance. For the fog-cloud environment, we propose a deep reinforcement learning-based task scheduling approach (DRLTS). In the proposed method, tasks are prioritized using fuzzy logic, categorized into high- and low-priority categories. Cloud nodes handle low-priority tasks, while fog nodes handle high-priority tasks. Deep Q-Networks (DQNs) are used to determine the optimal node at each level for task execution. Response time and energy consumption are minimized while task deadlines are met with our approach. According to our experiments, our approach outperforms heuristics, metaheuristics, and reinforcement learning-based scheduling techniques when it comes to response time, energy efficiency, deadline satisfaction, and performance overall.</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

Intelligent task scheduling for IoT in fog-cloud computing: a deep reinforcement learning approach

  • 期刊论文

  • 发表日期：

    2026年7月24日

查看全文
  • Reyhane Ghafari,
  • Najme Mansouri

摘要

The growth of IoT applications increases the demand for scalable computing resources. While cloud computing provides high processing and storage capacity, its centralized nature can cause latency and network congestion. Fog computing mitigates these issues by bringing computation closer to IoT devices, but its limited resources require efficient task scheduling to optimize performance. For the fog-cloud environment, we propose a deep reinforcement learning-based task scheduling approach (DRLTS). In the proposed method, tasks are prioritized using fuzzy logic, categorized into high- and low-priority categories. Cloud nodes handle low-priority tasks, while fog nodes handle high-priority tasks. Deep Q-Networks (DQNs) are used to determine the optimal node at each level for task execution. Response time and energy consumption are minimized while task deadlines are met with our approach. According to our experiments, our approach outperforms heuristics, metaheuristics, and reinforcement learning-based scheduling techniques when it comes to response time, energy efficiency, deadline satisfaction, and performance overall.