Scholarlink

<p>The increasing demand for real-time and energy-efficient task execution in Internet of Things (IoT) applications has positioned fog computing as a latency-sensitive alternative to traditional cloud-based processing. However, existing task-offloading strategies often neglect the multidimensional aspects of energy consumption—particularly parameters such as memory usage and data transfer—and rely on static heuristics that struggle in dynamic, heterogeneous environments. To address these limitations, we propose a new multi-objective energy-centric task offloading algorithm that integrates ECO-PSO and ECO-CSA with a dynamic coefficient adjustment mechanism based on gradient ascent. Our proposed solution explicitly models energy consumption in computing, transmission, and memory, along with makespan, enabling fine-grained task distribution among heterogeneous fog nodes. Furthermore, a customised energy model is introduced to select appropriate nodes with low CPU utilisation, enhancing energy efficiency and system performance. Numerous simulations conducted using the LEAF simulator in realistic fog environments demonstrate that our proposed method consistently outperforms PSG, PSG-M, CCFO, and MoAOA algorithms, achieving up to 25% lower total energy consumption and up to 30% reduction in makespan, while ensuring a more balanced task distribution. Therefore, the proposed algorithm is a valuable solution for creating future-oriented IoT systems in resource-constrained fog environments.</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

Multi-objective optimisation for energy-centric offloading in fog computing

  • 期刊论文

  • 发表日期：

    2025年10月24日

查看全文
  • Jaber Pournazari,
  • Amjad Ullah,
  • Ahmed Al-Dubai,
  • Xiaodong Liu,
  • Navid Khaledian

摘要

The increasing demand for real-time and energy-efficient task execution in Internet of Things (IoT) applications has positioned fog computing as a latency-sensitive alternative to traditional cloud-based processing. However, existing task-offloading strategies often neglect the multidimensional aspects of energy consumption—particularly parameters such as memory usage and data transfer—and rely on static heuristics that struggle in dynamic, heterogeneous environments. To address these limitations, we propose a new multi-objective energy-centric task offloading algorithm that integrates ECO-PSO and ECO-CSA with a dynamic coefficient adjustment mechanism based on gradient ascent. Our proposed solution explicitly models energy consumption in computing, transmission, and memory, along with makespan, enabling fine-grained task distribution among heterogeneous fog nodes. Furthermore, a customised energy model is introduced to select appropriate nodes with low CPU utilisation, enhancing energy efficiency and system performance. Numerous simulations conducted using the LEAF simulator in realistic fog environments demonstrate that our proposed method consistently outperforms PSG, PSG-M, CCFO, and MoAOA algorithms, achieving up to 25% lower total energy consumption and up to 30% reduction in makespan, while ensuring a more balanced task distribution. Therefore, the proposed algorithm is a valuable solution for creating future-oriented IoT systems in resource-constrained fog environments.