CCAC Theses and Dissertations
Campus Access Only
All rights reserved. This publication is intended for use solely by faculty, students, and staff of Nova Southeastern University. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, now known or later developed, including but not limited to photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author or the publisher.
Date of Award
2026
Document Type
Dissertation - NSU Access Only
Degree Name
Doctor of Philosophy in Information Systems (DISS)
Advisor
Wei Li
Keywords
Bio-Inspired Algorithms; Cybersecurity; Internet of Things; Threat Detection
Abstract
An Abstract of a Dissertation Submitted to Nova Southeastern University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy Applying Bio-Inspired Algorithms in Feature Selection for Machine Learning-Based IoT Threat Detection by Edward Smith August 2026 The rapid proliferation of Internet of Things (IoT) devices has expanded the cyberattack surface while producing high-dimensional network traffic that is difficult to process on resource-constrained gateways. This quantitative experimental study examined whether bio-inspired feature selection could improve the security-efficiency balance of machine learning-based intrusion detection systems. The study benchmarked 11 filter, wrapper, embedded, and swarm intelligence techniques and validated five leading methods across IoT-23, CIC-IDS2017, X-IIoTID, and N-BaIoT. Random Forest models were used to evaluate weighted F1-score, class-level detection behavior, training latency, processor utilization, and memory consumption. The results showed that no single feature selection technique dominated every environment. Ant Colony Optimization produced the strongest balance on transactional IoT-23 traffic, Lasso remained effective on standardized enterprise flows, and Boruta and Grey Wolf Optimization provided competitive alternatives for industrial and resource-constrained settings. Performance on N-BaIoT demonstrated that temporally aggregated streaming features create an architectural boundary that feature selection alone does not eliminate. The dissertation contributes a reproducible framework for evaluating IoT intrusion detection systems, a comparative assessment of feature engineering techniques, identification of domain-relevant feature subsets, empirical validation across heterogeneous datasets, and practical selection guidelines for researchers and information technology practitioners. The implementation is publicly available through a source code repository to support replication and extension of the research.
NSUWorks Citation
Edward Smith. 2026. Applying Bio-Inspired Algorithms in Feature Selection for Machine Learning-Based IoT Threat Detection. Doctoral dissertation. Nova Southeastern University. Retrieved from NSUWorks, . (1223)
https://nsuworks.nova.edu/gscis_etd/1223.