Adaptive Multi-Objective Task Scheduling in Cloud Computing Using Deep Reinforcement Learning and Hybrid Metaheuristic Optimization
Revolutionary Advances in Computing and Electronics: An International Journal · 12.06.2026 · 2 (1) · 57-78
DOI: 10.65890/race.v2i1.194 ↗Kurz gefasst
Fachartikel in der Zeitschrift Revolutionary Advances in Computing and Electronics: An International Journal, online erschienen am 12.06.2026. Verfasst von Saquib Ali und 2 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The rapid development of cloud computing has posed a significant challenge for task scheduling, given the highly dynamic, diverse, and large-scale workloads. Most heuristic and metaheuristic scheduling algorithms fail to adapt to rapidly changing run-time conditions, and lead to sub-optimal resource utilization, longer makespan, higher energy usage and violation of Service Level Agreements (SLAs). This study proposes an adaptive multi-objective task scheduling system based on Deep Reinforcement Learning (DRL) combined with hybrid metaheuristic optimisation methods to overcome these restrictions. To enable independent decision-making in an unpredictable, fluctuating workload, the proposed approach dynamically learns optimal scheduling rules from real-time conditions in the cloud system using a Deep Q-Network (DQN). In addition, the integration of Genetic Algorithm and Particle Swarm Optimization (PSO) is performed to encourage convergence towards the optimal scheduling solution without the occurrence of local minima. The hybrid approach has been shown to find a good balance between exploration and exploitation in complex cloud environments. We further evaluate the practicality of the proposed framework through large-scale simulations of real-world cloud systems running Kubernetes and OpenStack with CloudSim. The performance is assessed based on key metrics such as makespan, resource utilisation, energy efficiency, throughput and SLA violation rate. Through experiments, it has been shown that the proposed method consistently outperforms the state-of-the-art scheduling strategies in execution time, energy efficiency, resource utilization, and fewer SLA violations. The study presents a scalable, intelligent, and self-adaptive scheduling paradigm suitable for effective, persistent resource management in extremely dynamic distributed computing environments for next-generation cloud infrastructures.
Zitieren
APA
Ali, S., Haidri, R. A., & Farooqui, N. A. (2026). Adaptive Multi-Objective Task Scheduling in Cloud Computing Using Deep Reinforcement Learning and Hybrid Metaheuristic Optimization. Revolutionary Advances in Computing and Electronics: An International Journal, 2(1), 57-78. https://doi.org/10.65890/race.v2i1.194
Vancouver
Ali S, Haidri RA, Farooqui NA. Adaptive Multi-Objective Task Scheduling in Cloud Computing Using Deep Reinforcement Learning and Hybrid Metaheuristic Optimization. Revolutionary Advances in Computing and Electronics: An International Journal. 2026;2(1):57-78. doi:10.65890/race.v2i1.194
BibTeX
@article{Ali2026,
author = {Ali, Saquib and Haidri, Raza Abbas and Farooqui, Nafees Akhter},
title = {Adaptive Multi-Objective Task Scheduling in Cloud Computing Using Deep Reinforcement Learning and Hybrid Metaheuristic Optimization},
journal = {Revolutionary Advances in Computing and Electronics: An International Journal},
year = {2026},
volume = {2},
number = {1},
pages = {57--78},
doi = {10.65890/race.v2i1.194}
}
EUVIMED