Electronic Thesis/Dissertation
 

IoT Attack Detection Using a Stacked Ensemble for Multi-Class Classification

Open Access Deposited

The rapid expansion of the Internet of Things (IoT) has introduced a vast attacksurface, with constrained devices often lacking the security needed to withstand modern multi-vector threats. This praxis proposes and evaluates a stacked ensemble intrusiondetection framework that combines a Convolutional Neural Network (CNN), a Long Short-Term Memory network (LSTM), and a LightGBM gradient-boosted tree to detect and classify 21 IoT attacks types using the CICIoT2023 dataset. Following three stage exploratory data analysis and thorough preprocessing, the individual models were trained and their predictions fused via a second level LightGBM meta-learner. On unseen test data, the CNN achieved an F1 of 0.96, the LSTM had an F1 of 0.93, and the LightGBM produced a F1 of 0.92. The stacked ensemble further improved the F1 to 0.98. These results validate a practical, high-accuracy, and resource-efficient solution for real-time, multi-class IoT threat detection, paving the way for secure deployment on edge gateways.

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 Washington_gwu_0075A_17558.pdf Washington_gwu_0075A_17558.pdf 2025-12-11 Open Access