Scholarlink

<p>As computational tasks are increasingly processed on edge devices, resource allocation and task scheduling have been shifted from centralized cloud platforms to decentralized edge clusters. However, task scheduling in resource-constrained edge environments is confronted with significant challenges in resource coordination and load balancing due to limited resources, heterogeneous hardware configurations, and diverse task requirements. To address these challenges, a multi-resource-aware load balancing scheduling strategy for heterogeneous edge clusters is proposed in this paper. The strategy dynamically monitors the residual capacities of multiple resources–including CPU, memory, disk I/O, and network I/O–across heterogeneous nodes and matches these capacities with the diverse resource demands of incoming tasks. Adaptive resource allocation and efficient task scheduling are thereby enabled, resulting in improved overall scheduling quality. To support the optimized execution of this scheduling strategy, a scheduling-driven enhanced search algorithm is designed that integrates a staged adaptive weight mechanism, a perturbation-based jump search strategy, and a cooperative operator mechanism to enhance global search capability and optimization accuracy during the scheduling process. Experimental evaluations were conducted on a real-world heterogeneous edge cluster across four representative task scenarios. In comparison with existing methods, the proposed strategy not only improves resource utilization efficiency but also significantly enhances cluster load balancing, thereby demonstrating superior adaptability and scheduling performance.</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

A Multi-Resource-Aware and Load-Balanced Scheduling Strategy for Heterogeneous Edge Clusters

  • 期刊论文

  • 发表日期：

    2025年12月8日

查看全文
  • Xiang Li,
  • Enguo Zhu,
  • Jian Qin,
  • Weiyang Xu,
  • Yucheng Wu

摘要

As computational tasks are increasingly processed on edge devices, resource allocation and task scheduling have been shifted from centralized cloud platforms to decentralized edge clusters. However, task scheduling in resource-constrained edge environments is confronted with significant challenges in resource coordination and load balancing due to limited resources, heterogeneous hardware configurations, and diverse task requirements. To address these challenges, a multi-resource-aware load balancing scheduling strategy for heterogeneous edge clusters is proposed in this paper. The strategy dynamically monitors the residual capacities of multiple resources–including CPU, memory, disk I/O, and network I/O–across heterogeneous nodes and matches these capacities with the diverse resource demands of incoming tasks. Adaptive resource allocation and efficient task scheduling are thereby enabled, resulting in improved overall scheduling quality. To support the optimized execution of this scheduling strategy, a scheduling-driven enhanced search algorithm is designed that integrates a staged adaptive weight mechanism, a perturbation-based jump search strategy, and a cooperative operator mechanism to enhance global search capability and optimization accuracy during the scheduling process. Experimental evaluations were conducted on a real-world heterogeneous edge cluster across four representative task scenarios. In comparison with existing methods, the proposed strategy not only improves resource utilization efficiency but also significantly enhances cluster load balancing, thereby demonstrating superior adaptability and scheduling performance.