Public reproducibility artifact for the PhD dissertation:
Phornsawan Roemsri. Data-Driven Cybersecurity for Connected Devices: Traffic, Provenance, and Behavioral Threat Detection. PhD dissertation, Texas Tech University, December 2026.
This repository contains the code, data preparation scripts, and configurations for Chapters 2 through 5, so that each comparison in the dissertation can be rerun.
| Folder | Chapter | What it reproduces |
|---|---|---|
| ch2-device-identification | 2 | ML classifiers vs. a modified Jaccard similarity baseline for IoT device identification (677 real + 12,000 synthetic protocol records) |
| ch3-provenance-authentication | 3 | Provenance/location-based authentication policy vs. a decision tree (5,000 synthesized requests) |
| ch4-ransomware-detection | 4 | Threshold-based ransomware detection from CPU/memory/disk/I/O monitoring (Dstat data) |
| ch5-attack-visualization | 5 | 3D visualizations of flooding attacks from the CIC IoT Dataset 2022 |
Each chapter folder has its own README describing exactly what it
reproduces and how to run it, with the code in code/, dataset
preparation scripts in data-prep/, and run configurations in config/.
Datasets whose redistribution rights are unclear are not included in this repository. In particular, the CIC IoT Dataset 2022 must be obtained from its original source (see the Chapter 5 README); this repository provides only the preparation scripts that transform the original download into the inputs used by the experiments. Small synthesized or derived datasets generated by the scripts here are reproducible from the scripts themselves.
python -m venv .venv
# Windows: .venv\Scripts\activate POSIX: source .venv/bin/activate
pip install -r requirements.txtThen follow the "How to run" section of the chapter you want to reproduce.
All four chapters (2–5) are complete and verified against the dissertation. Each chapter's README describes exactly what it reproduces, any known limitations, and how to run it.
This artifact is archived on Zenodo. Cite it via the concept DOI 10.5281/zenodo.24132545, which always resolves to the latest version. Citation metadata is in CITATION.cff.