Date of Graduation
Fall 2025
Degree
Master of Science in Computer Science
Department
Computer Science
Committee Chair
Ajay Katangur
Abstract
Cloud computing has grown rapidly in recent years, mainly due to the sharp increase in data transferred over the internet. This growth makes task scheduling a key and challenging part of cloud systems, as it helps distribute user requests across servers to minimize response time, prevent overloading, and ensure smooth user experience. This thesis proposes two novel approaches for dynamic task scheduling in cloud environments. First, a novel Score-Based Dynamic Load Balancing (SBDLB) strategy is developed, which leverages system parameters to allocate tasks efficiently across virtual machines (VMs) in data centers. SBDLB ensures balanced workload distribution by continuously evaluating VM suitability, minimizing response time and preventing resource bottlenecks. Second, this thesis also introduces a Deep Reinforcement Learning (DRL) based framework using a Double Deep Q-Network (DDQN), where the agent learns optimal task allocation policies through system state data and a reward structure that promotes efficient resource use, balanced load distribution, and high overall performance. The effectiveness of both approaches was extensively evaluated using the CloudSim 7G platform. For SBDLB, experiments across diverse workloads demonstrated significant improvements over the throttled load balancing strategy. Similarly, the RL-based scheduler consistently outperformed traditional baselines, including Throttled and SBDLB, under varying workload conditions, reducing average response time up to 134% over throttled and 67% over SBDLB. When deployed in a resource-constrained environment with a 20% reduction in available resources, the pre-trained model maintained comparable performance with minimal fine-tuning, underscoring its adaptability in fault-prone or limited-resource scenarios. Together, these contributions offer scalable, effective scheduling strategies that can improve real-world cloud environments and help drive future advancements in the field.
Keywords
cloud computing, dynamic load balancing, reinforcement learning, DDQN, task scheduling
Subject Categories
Artificial Intelligence and Robotics | Other Computer Sciences
Copyright
© Shadman Sakib
Recommended Citation
Sakib, Shadman, "Efficient Task Scheduling in Cloud Infrastructures Using Dynamic Score-Based Allocation and Deep Q-Learning" (2025). Graduate Theses/Dissertations. 4118.
https://bearworks.missouristate.edu/theses/4118
Open Access
