Scholarlink

<p>As scientific focus shifts toward sustainability, research increasingly integrates eco-friendly solutions with technological innovation to reduce carbon emissions and enhance planetary habitability. Therefore, this paper highlights those sustainable solutions which are involve improving the energy efficiency of cloud data centres by using deep reinforcement learning. Many such strategies are proposed by many researchers across the globe and different reports and studies are available claiming for better energy efficient solution in high workload environment. Green Cloud Computing (GCC) maximizes the usage of energy while maintaining service-level quality. Deep Reinforcement Learning (DRL) has been an efficient paradigm for smart and adaptive resource allocation in datacentres over the last few years. With a focus on workload scheduling, dynamic resource provisioning, cooling control, and energy-efficient VM placement, this paper presents a detailed overview of DRL techniques’ utilization in the GCC. We carry out a thorough analysis of advanced DRL algorithms, such as hybrid MILP-RL methods, Proximal Policy Optimization (PPO), Multi-Agent DRL (MADRL), and Deep Q-Networks (DQN). Performance metrics including energy consumption, SLA adherence, makespan, and throughput are employed to evaluate the works that have been assessed. Existing gaps in research are illustrated in this paper and paths towards the creation of intelligent, scalable, and sustainable cloud systems are proposed.</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 survey of deep reinforcement learning techniques for Energy-efficient green cloud computing

  • 期刊论文

  • 发表日期：

    2025年10月9日

查看全文
  • Akhilesh Chauhan,
  • Neeraj Kumar Pandey,
  • Manoj Diwakar,
  • Prabhishek Singh,
  • Amit Kumar Mishra

摘要

As scientific focus shifts toward sustainability, research increasingly integrates eco-friendly solutions with technological innovation to reduce carbon emissions and enhance planetary habitability. Therefore, this paper highlights those sustainable solutions which are involve improving the energy efficiency of cloud data centres by using deep reinforcement learning. Many such strategies are proposed by many researchers across the globe and different reports and studies are available claiming for better energy efficient solution in high workload environment. Green Cloud Computing (GCC) maximizes the usage of energy while maintaining service-level quality. Deep Reinforcement Learning (DRL) has been an efficient paradigm for smart and adaptive resource allocation in datacentres over the last few years. With a focus on workload scheduling, dynamic resource provisioning, cooling control, and energy-efficient VM placement, this paper presents a detailed overview of DRL techniques’ utilization in the GCC. We carry out a thorough analysis of advanced DRL algorithms, such as hybrid MILP-RL methods, Proximal Policy Optimization (PPO), Multi-Agent DRL (MADRL), and Deep Q-Networks (DQN). Performance metrics including energy consumption, SLA adherence, makespan, and throughput are employed to evaluate the works that have been assessed. Existing gaps in research are illustrated in this paper and paths towards the creation of intelligent, scalable, and sustainable cloud systems are proposed.