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