Scholarlink

<p>This paper proposes the Adaptive Memetic-Guided Slime Evolution Algorithm (AMGSEA) for multi-objective task scheduling in large-scale Infrastructure-as-a-Service (IaaS) cloud environments. Unlike conventional hybrid metaheuristics that rely on static operator integration, AMGSEA introduces a feedback-driven adaptive framework that dynamically balances exploration and exploitation. The proposed method combines oscillatory global search from the Slime Mould Algorithm, Differential Evolution-based adaptive guidance, and a selective memetic local search applied only to elite non-dominated solutions. The key novelty lies in (i) adaptive activation of memetic refinement based on Pareto dominance, (ii) feedback-controlled evolutionary guidance to prevent premature convergence, and (iii) an elite re-injection strategy for diversity preservation. Extensive experiments using CloudSim with PlanetLab traces and synthetic workloads of up to 5,000 tasks demonstrate that AMGSEA achieves up to <b>14.6% reduction in makespan</b>, <b>11.2% reduction in execution cost</b>, and improved energy efficiency compared to seven state-of-the-art schedulers. Additionally, the method improves deadline satisfaction to <b>97% compliance</b> and increases hypervolume by an average of <b>8–12%</b>. Statistical validation using Wilcoxon signed-rank tests confirms the significance of the improvements. These results establish AMGSEA as a scalable, QoS-aware, and energy-efficient scheduling framework suitable for dynamic and heterogeneous cloud 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

Adaptive Memetic Guided Slime Evolution Algorithm for Multi-Objective Task Scheduling in Large-Scale Cloud Environments

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月18日

查看全文
  • R. Nithiavathy,
  • P. M. Benson Mansingh,
  • R. Nallakumar,
  • A. Britto Manoj

摘要

This paper proposes the Adaptive Memetic-Guided Slime Evolution Algorithm (AMGSEA) for multi-objective task scheduling in large-scale Infrastructure-as-a-Service (IaaS) cloud environments. Unlike conventional hybrid metaheuristics that rely on static operator integration, AMGSEA introduces a feedback-driven adaptive framework that dynamically balances exploration and exploitation. The proposed method combines oscillatory global search from the Slime Mould Algorithm, Differential Evolution-based adaptive guidance, and a selective memetic local search applied only to elite non-dominated solutions. The key novelty lies in (i) adaptive activation of memetic refinement based on Pareto dominance, (ii) feedback-controlled evolutionary guidance to prevent premature convergence, and (iii) an elite re-injection strategy for diversity preservation. Extensive experiments using CloudSim with PlanetLab traces and synthetic workloads of up to 5,000 tasks demonstrate that AMGSEA achieves up to 14.6% reduction in makespan, 11.2% reduction in execution cost, and improved energy efficiency compared to seven state-of-the-art schedulers. Additionally, the method improves deadline satisfaction to 97% compliance and increases hypervolume by an average of 8–12%. Statistical validation using Wilcoxon signed-rank tests confirms the significance of the improvements. These results establish AMGSEA as a scalable, QoS-aware, and energy-efficient scheduling framework suitable for dynamic and heterogeneous cloud environments.