Electronic Thesis/Dissertation
 

Federated Learning Threat Classification with Heterogeneous Devices in Converged Energy Sector Networks

Open Access Deposited

From Edge to Enterprise

Cyberattacks targeting the U.S. Energy Sector's converged networks areescalating in frequency and sophistication and present a significant risk that could cost the economy over $1 trillion in loses. The current threat landscape creates an urgent need for innovative, scalable, and adaptable threat detection systems that can operate across diverse hardware. To address this challenge, our research proposes and evaluates a Heterogeneous Federated Multi-layer Perceptron Threat Classification (HFMLP-TC) model. This approach utilizes a TensorFlow federated learning framework that uniquely incorporates computationally diverse clients, a high-performance i9 workstation and a resource constrained Raspberry Pi 5, for collaborative threat detection. Results demonstrate that the HFMLP-TC model significantly outperformed prior benchmark research, achieving a detection rate exceeding 99.9% and reducing the false negative rate to 0.03%. The results of our study lend credibility to the idea that integrating computationally constrained devices into real‑world cyber defense operations is not just a theoretical possibility but a practical one.

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 Soto_gwu_0075A_17469.pdf Soto_gwu_0075A_17469.pdf 2025-12-11 Open Access