Jul 2026· Journal of Science Research and Reviews· 0 citations· 1 references
TL;DR
Experimental results demonstrate that DL-EATS achieves the lowest energy consumption, shortest makespan, minimal SLA violation rate, and highest resource utilization, representing an 18.5% improvement in energy efficiency over the next best method and substantial gains across all performance metrics.
Abstract
Energy efficiency in cloud computing has become a critical concern due to the growing energy demands of large-scale datacenters, operational costs and environmental impacts. This research proposes Deep Learning-Enhanced Energy-Aware Task Scheduling (DL-EATS), a novel solution that combines LSTM-based workload prediction with a hybrid multi-objective optimization model integrating Genetic Algorithm and Particle Swarm refinement. DL-EATS intelligently schedules tasks to virtual machines or containers by anticipating workload variations, thereby minimizing energy consumption, ensuring SLA compliance and maximizing system throughput. The method was evaluated against state-of-the-art algorithms, including Reinforcement Learning-based Virtual Machine Placement (RLVMP), Enhanced Adaptive Moth-Flame Optimization (EA-MFO), Q-Learning-based Resource Dynamic Optimization (Q-RDO) and Task Scheduling using Grey Wolf Optimizer (TS-GWO), under identical cloud workload scenarios. Experimental results demonstrate that DL-EATS achieves the lowest energy consumption (480 kWh), shortest makespan (38 s), minimal SLA violation rate (1.2%) and highest resource utilization (92%), representing an 18.5% improvement in energy efficiency over the next best method and substantial gains across all performance metrics. These findings confirm that integrating predictive deep learning with hybrid heuristic optimization provides a scalable, reliable and energy-efficient solution for modern cloud datacenter management.
An AI-enabled dynamic task scheduling framework based on Deep Reinforcement Learning (DRL) with a Deep Q-Network (DQN) model to dynamically assign tasks to virtual machines and learn the best scheduling policies by continuously interacting with the cloud environment based on system parameters such as resource availability, task queue length, and virtual machine load is introduced.
Karnam Sreenu, G. Prasadu, K. Premnadh et al.· VFAST Transactions on Softwa...· 0 citations
Energy-aware task scheduling in heterogeneous cloud infrastructures remains challenging due to the combinatorial growth of task-to-resource assignments, resource heterogeneity, and the need to balance energy consumption with scheduling performance. This paper proposes an Adaptive Dominance-Guided Grey Wolf Optimizer (ADG-GWO) for non-preemptive task scheduling in heterogeneous cloud environments. ADG-GWO adapts Grey Wolf Optimization to discrete task-to-VM assignment by integrating dominance-guided genetic reproduction, Hamming-distance-based diversity regulation, and adaptive reproduction control. These mechanisms are designed to improve search stability, reduce premature convergence, and support effective exploration of high-dimensional assignment spaces without expanding the externally tuned hyperparameter space.The proposed method is evaluated through simulation under workload-scaling and capacity-scaling scenarios using heterogeneous cloud configurations. For evaluation, workload instances and heterogeneous VM configurations are derived from Google Cluster Trace 2019 to instantiate realistic task-to-VM scheduling scenarios. The results show that the proposed dominance-guided adaptive search improves energy-aware scheduling effectiveness while maintaining competitive scheduling efficiency in heterogeneous cloud environments.
Saleh Al Shamaa, Wei Shi, J. Corriveau· IEEE International Conferenc...· 0 citations
Minimizing energy consumption and carbon emissions while maintaining system performance is a critical challenge in cloud task scheduling. This paper presents a multi-objective scheduling framework based on a Memetic Algorithm (MA) designed to optimize task-to-VM mapping with respect to energy efficiency, carbon footprint, and throughput. The algorithm employs a weighted fitness function that integrates actual and idle energy usage, simulated time-varying carbon intensity, and task throughput. To enhance solution quality, MA combines global evolutionary operations (selection, crossover, mutation) with local search heuristics that adaptively refine candidate solutions based on workload characteristics and green energy opportunities. The carbon emission model incorporates dynamic emission factors (γ) derived from location- and time-sensitive datasets, reflecting real-world variability in grid carbon intensity. The proposed method is evaluated using the NASA Ames iPSC/860 workload under both low and high resource utilization scenarios. Comparative results demonstrate that the proposed MA approach achieves reduces the carbon emission by 20.2%, minimizes energy consumption by 17.7%, and enhances throughput by 21.2% over conventional techniques such as HDDPGTS and RAPTS, while also ensuring competitive performance in terms of makespan and resource utilization. These improvements underscore the potential of memetic-based hybrid scheduling to support environmentally sustainable and performance-efficient cloud infrastructures. The findings highlight the importance of integrating eco-aware intelligence into task scheduling policies, particularly for mission-critical and energy-intensive cloud applications.
Chennoji Sandhya, Mandla Alphonsa, Vankudoth Biksham et al.· International Journal of Int...· 0 citations
Due to the rapid development of cloud computing services, today's data centers are consuming much more energy and emitting much higher levels of carbon dioxide into the atmosphere compared to earlier years, thus requiring the development of sustainable solutions for resource management in the cloud. In this paper, an approach called Carbon Aware Cloud Resource Scheduling based on the use of Reinforcement Learning is presented. The proposed approach includes workload demand, resource utilization, and carbon intensity values in its decision making process in order to allocate workloads efficiently and minimize environmental impact caused by cloud computing services. The reinforcement learning algorithm trains the RL agent to find the optimal strategies for workload scheduling that involve execution of workloads, postponing their execution, migrating them from one location to another and re-allocation of resources. The results from experimental evaluation indicate that the suggested framework successfully reduces the carbon emission level up to 30% and energy cost level to around 24%, respectively, in comparison with traditional scheduling methods, without lowering the SLA compliance rate down to 98.4% and workload starvation level up to 0.3%. The analysis proves that the combination of carbon awareness and reinforcement learning helps to create an intelligent, adaptive, and ecologically sustainable system for managing cloud resources.
Kirupavathy P., Hareeni C., Jayashri K. et al.· Journal of Ubiquitous Comput...· 0 citations
Cloud computing enables on-demand access to scalable virtualized resources. However, efficient task scheduling in cloud computing remains a challenge because of the dynamic and heterogeneous nature of workloads. This paper proposes a hybrid Deep Reinforcement Learning (DRL) framework that combines Deep Q-Network (DQN), Proximal Policy Optimization (PPO) and Advantage Actor-Critic (A2C) to enable adaptive resource scheduling in cloud environments. The proposed model is implemented using PyTorch and evaluated in a CloudSim based simulation environment Experimental results show that the proposed approach achieves a consistent improvement in terms of makespan reduction and VM utilization compared to individual DRL approaches and classical scheduling algorithms. Experiments were reiterated with multiple runs to ensure reliability and statistical measures are reported. Under the evaluated conditions, the proposed approach shows more efficient scheduling performance, but it has higher computational overhead and is only validated in a simulated environment for now. These results suggest that hybrid DRL-based scheduling is a promising approach for adaptive cloud resource management, with potential for further validation in real-world deployments and energy-aware scenarios.
P. Priya, J. Geetha, E. Naresh et al.· International Journal of Com...· 0 citations
Cloud computing has transformed the delivery of modern applications and services by providing scalable, flexible, and cost-effective access to computing resources. One of the most critical challenges in cloud environments is the efficient distribution of dynamic workloads across heterogeneous resources, commonly addressed through load balancing and task scheduling techniques. Efficient scheduling plays a vital role in maximizing resource utilization, minimizing response time, and maintaining acceptable Quality of Service (QoS), particularly under dynamic and large-scale workloads. Despite the progress achieved by traditional heuristics such as Min-Min and metaheuristic approaches like the Improved Sparrow Search Algorithm (ISSA), challenges related to scalability, adaptability, and computational overhead remain. Metaheuristic-based approaches often involve iterative optimization processes that may limit their efficiency in real-time scheduling scenarios. In this paper, we propose a lightweight Stochastic Predictive Energy-Aware Scheduling (SPES) algorithm that integrates predictive execution estimation, multi-resource awareness, and stochastic decision-making. Unlike deterministic scheduling strategies, SPES employs a Top K candidate selection mechanism combined with probabilistic weighting and epsilon-greedy exploration to enhance adaptability and avoid suboptimal resource allocation. The proposed method considers CPU, memory, and I/O demands to achieve balanced utilization across heterogeneous hosts while implicitly addressing energy efficiency through utilization-based modeling. The proposed algorithm is implemented and evaluated using the CloudSim 5.0 simulation framework under heterogeneous multi-region cloud environments with varying workload sizes. Experimental results demonstrate that SPES consistently outperforms ISSA and achieves makespan reductions of up to 23.8% while improving scalability, resource utilization, and scheduling efficiency under dynamic cloud workloads. These results indicate that SPES provides an effective lightweight scheduling solution for large-scale and energy-aware cloud computing environments and supports green computing objectives through improved resource efficiency.
M. Yacoub, A. Raouf, Walaa K. Gad et al.· Electronics· 0 citations