Theses and Dissertations

ORCID

https://orcid.org/0009-0000-7134-7405

Issuing Body

Mississippi State University

Advisor

Khan, Samee

Committee Member

Malik, Asad Waqar

Committee Member

Jones, Bryan

Committee Member

Luo, Yu

Date of Degree

8-7-2025

Original embargo terms

Immediate Worldwide Access

Document Type

Dissertation - Open Access

Major

Electrical & Computer Eng

Degree Name

Doctor of Philosophy (Ph.D.)

College

James Worth Bagley College of Engineering

Department

Department of Electrical and Computer Engineering

Abstract

Unlike cloud-based systems, where resources are centralized for data processing and located at a distance from the source, leading to an increase in the response time, edge computing emerges as a transformative paradigm that processes the data closer to the sources to meet the application’s service requirements. Since the resources in edge computing are distributed and limited, load balancing is crucial. To ensure the prevention of bottlenecks and failures in resource-constrained environments, load balancing is an important aspect of edge computing as it optimizes resource utilization while maintaining the requirements of low-latency applications, improved fault tolerance, energy efficiency, adaptability, scalability, and ensures reliability by efficiently distributing the network load. This dissertation explores the different techniques of balancing load for stream processing in edge computing through three phases; each phase focuses on optimizing the edge performance by balancing the load for a resource-constrained environment. In the first phase of our dissertation research, we focus on efficient data processing and performance modeling. In this phase, we develop a framework for the intelligent scheduling of tasks based on the task priority. In this phase, we also probe the probabilistic methods to analyze and predict system performance under uncertain workload conditions. In the second phase, we delve into data management and data acquisition techniques for edge computing systems. In this phase, we worked on providing the load balancing solution using the capabilities of machine learning and artificial intelligence. Firstly, in this round, we propose an intelligent machine learning based aggregation node selection framework to mitigate the impact of congestion in a resource-constrained environment. During this phase, we also developed a framework for intelligent Generative AI management (iGenEdge), designed for IoT edge devices, which aims to dynamically provision resources to handle varying demands. Finally, in the last phase, we perform system benchmarking and performance evaluation using a testbed and virtual machines. By addressing these key challenges in load balancing for edge-based stream processing, this dissertation contributes to the development of scalable, energy-efficient, and fault-tolerant solutions that enhance the reliability and performance of edge computing systems. The findings provide valuable insights for optimizing distributed computing architectures and advancing real-time data processing in resource-constrained environments.

Download

Included in

Engineering Commons

Share

COinS