Electronic Thesis/Dissertation
 

Machine Learning Approach to Predicting IoT Cybersecurity Attacks in U.S. MSPs Remote Work Scenarios

Open Access Deposited

In the evolving landscape of cybersecurity, within the perspective of Internet of Things (IoT) devices within remote work environments, machine learning (ML) models present a formidable defense mechanism. This praxis evaluates the effectiveness of supervised ML models—Random Forest, XGBoost, and Multiclass Classification Artificial Neural Networks—in predicting and classifying cybersecurity threats posed to U.S. Managed Service Providers (MSPs). Utilizing extensive datasets, NF-UQ-NIDS-v2 and CICIDS2017, the study demonstrates that ML models can achieve high accuracy and maintain low false positive rates, confirming the hypothesis that a limited set of key features is sufficient for effective threat prediction. The research underscores the strategic importance of feature selection, model tuning, and the innovative application of the Cyber Kill Chain framework to enhance threat detection and classification. Findings suggest that Random Forest is particularly effective for environments demanding high precision with minimal false alarms, while XGBoost excels in scenarios where capturing a broad range of threats is critical. The study advances the body of knowledge providing a novel categorization method with empirical evidence on the capabilities of ML in securing access to critical infrastructures against advanced threats, offering recommendations for integrating these technologies within existing cybersecurity frameworks.

Author Language Keyword Date created Type of Work License
  • All rights reserved
Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL

Notice to Authors

If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.

Thumbnail Title Date Uploaded Visibility Actions
Preview of Johnstone_gwu_0075A_16980.pdf Johnstone_gwu_0075A_16980.pdf 2025-04-09 Open Access