AI-Based Dynamic Task Scheduling in Cloud Computing Using Deep Reinforcement Learning

Abstract

Effective task scheduling is essential in cloud computing to maximize resource utilization and satisfy Service Level Agreement (SLA) requirements in highly dynamic and heterogeneous environments. Conventional scheduling algorithms, like Round Robin, Min-Min, and metaheuristic-based methods, rely on static or preset rules that are unable to respond to changes in the workload in real time. This results in longer execution times, higher energy consumption, and less-than-ideal system performance. In order to overcome these constraints, this paper introduces 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. To balance makespan, energy consumption, throughput, and SLA violation rate, a multi-objective reward function is developed. The proposed DRL-based scheduler for intelligent and adaptive scheduling in present cloud computing settings is deployed and tested in CloudSim Plus simulator platform for diverse workload situations to affirm the usefulness of the framework. The results demonstrate that the DRL-based scheduler outperforms the conventional scheduling schemes by 23.6% in makespan, 27% in energy efficiency, 49% in SLA violations, and 33% in throughput.