Scholarlink

The ability to compute all possible mappings between workflow tasks and IaaS cloud resources, to identify an optimal configuration, is computationally infeasible. The security of all precedence-constraints tasks in a public IaaS cloud introduces a critical dimension to the scheduling problem, particularly for sensitive applications such as healthcare, finance, and defence. This study presents a hybrid scheduling approach, Q-learning-enhanced Firefly Algorithm (Q-FA), to optimize makespan and security of precedence-constraints workflow tasks in an IaaS cloud environment. The proposed Q-FA algorithm adapts mutation actions using reinforcement learning, enabling dynamic task scheduling based on reward-driven decisions. Experimental results on DAG-based workflows demonstrated that the proposed Q-FA outperforms Firefly Algorithm (FA), Whale Optimization Algorithm (WOA), Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and Salp Swarm Algorithm (SSA) in minimizing makespan and security levels achievement. The convergence analysis of Q-FA highlighted the fast and stable optimization behavior.

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

Enhancing Scheduling with Q-Learning and Firefly Algorithm in IaaS Cloud Environment: A Security-Aware Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zeya Mustafa,
  • Mohammad Qasim,
  • Mohammad Sajid

摘要

The ability to compute all possible mappings between workflow tasks and IaaS cloud resources, to identify an optimal configuration, is computationally infeasible. The security of all precedence-constraints tasks in a public IaaS cloud introduces a critical dimension to the scheduling problem, particularly for sensitive applications such as healthcare, finance, and defence. This study presents a hybrid scheduling approach, Q-learning-enhanced Firefly Algorithm (Q-FA), to optimize makespan and security of precedence-constraints workflow tasks in an IaaS cloud environment. The proposed Q-FA algorithm adapts mutation actions using reinforcement learning, enabling dynamic task scheduling based on reward-driven decisions. Experimental results on DAG-based workflows demonstrated that the proposed Q-FA outperforms Firefly Algorithm (FA), Whale Optimization Algorithm (WOA), Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and Salp Swarm Algorithm (SSA) in minimizing makespan and security levels achievement. The convergence analysis of Q-FA highlighted the fast and stable optimization behavior.