Resource-Efficient Edge Computing and Lightweight Traffic Fingerprinting for Scientific Applications

Thumbnail Image

Authors

Rouf, Abdur

Issue Date

2024

Type

Thesis

Language

en_US

Keywords

Edge Computing , Genetic Algorithm Optimization , IoT Device Identification , Locality-Sensitive Hashing , Network Traffic Fingerprinting , Resource-Constrained Environments

Research Projects

Organizational Units

Journal Issue

Alternative Title

Abstract

The explosive growth of the Internet of Things (IoT) and the increasing demand for real-time, data-intensive applications present significant challenges in network management, device identification, and resource efficiency. This thesis addresses these challenges through two approaches: resource-efficient edge computing for scientific applications and lightweight network traffic fingerprinting for IoT environments.In fire detection systems like AlertWildfire, integrating edge devices significantly reduces latency by up to 70% and bandwidth consumption by 51% by enabling local data processing. For IoT device identification, this research introduces FlexiNet, a lightweight fingerprinting system using genetic algorithms, achieving 96% accuracy while reducing processing time and memory usage by 7% and 56%, respectively. An alternative Locality-Sensitive Hashing (LSH) method further improves accuracy by 12%, demonstrating its scalability in constrained environments. These advancements highlight the synergy between edge computing and efficient traffic fingerprinting in addressing the unique demands of scientific and industrial IoT systems. By optimizing both latency-critical operations and resource-intensive identification processes, this research bridges the gap between computational efficiency and scalability while providing a flexible framework adaptable to broader IoT applications. The findings lay a foundation for enhancing network management, improving security, and fostering the development of adaptive systems in constrained and dynamic environments.

Description

Citation

Publisher

License

Journal

Volume

Issue

PubMed ID

DOI

ISSN

EISSN