Title: Multi-objective enhanced task scheduling algorithm for SLA violation mitigation in cloud computing using deep reinforcement learning

Authors: Mallu Shiva Rama Krishna; Khasim Vali Dudekula

Addresses: School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, India ' School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, India

Abstract: Task scheduling in distributed heterogeneous cloud environments presents challenges due to NP-hard complexity, non-linear dynamics, and multi-objective optimisation requirements. To address these challenges, we propose an intelligent scheduling framework that combines real-time workload monitoring with predictive task analysis using Deep Q-Network (DQN) reinforcement learning. Our adaptive solution continuously optimises resource allocation through experiential learning, improving system performance while reducing energy consumption and minimising SLA violations in dynamic cloud environments. We evaluate the Multi-Objective Enhanced Task Scheduling Algorithm for SLA Violation Mitigation in Cloud Computing using Deep Reinforcement Learning (METCD) against baseline algorithms, including WOA, CA, GWO, ALO, and GA. The results demonstrate that METCD outperforms existing approaches, reducing makespan by 28.76% to 40.03%, SLA violations by 29.59% to 47.12%, and energy consumption by 32.82% to 42.2%. These improvements emphasise the efficiency of METCD in optimising task scheduling within cloud computing environments.

Keywords: SLA violation; cloud computing; task scheduling; energy consumption; machine learning; makespan; deep Q-learning.

DOI: 10.1504/IJGUC.2025.148552

International Journal of Grid and Utility Computing, 2025 Vol.16 No.5/6, pp.588 - 603

Received: 11 Jun 2024
Accepted: 29 Jul 2024
Published online: 11 Sep 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article