IoTSecUT: Uncertainty-Based Hybrid Deep Learning Approach for IoT Security

📌 Overview

This repository contains the implementation of IoTSecUT, an uncertainty-based hybrid deep learning approach for securing IoT ecosystems against evolving cyber threats. The framework leverages:

• Conditional Generative Adversarial Network (cGAN) for minority class upsampling.
• Auxiliary Autoencoder for high-dimensional data reduction.
• Hybrid Uncertainty-Based Transformer Model for effective network traffic classification.

The research was published in IEEE IoT Journal and demonstrates the superiority of our approach in handling class imbalance, data dimensionality challenges, and efficient IoT security.

---

🚀 Features

• Robust intrusion detection using hybrid deep learning.
• Generative augmentation for class imbalance correction.
• Dimensionality reduction for optimizing network traffic data.
• Uncertainty estimation for improved classification reliability.
• Evaluated on real-world datasets: BoT-IoT & CICIDS2018.

---

📂 Dataset

We utilize two well-known intrusion detection datasets:

• BoT-IoT: Contains labeled network traffic data for IoT environments.
• CICIDS2018: Provides a realistic intrusion dataset for evaluating IDS models.

To obtain these datasets:

1. BoT-IoT: Download Here
2. CICIDS2018: Download Here

---

🛠 Installation

1️⃣ Clone the repository

$ git clone https://github.com/your-repo/iotsecut.git
$ cd iotsecut

2️⃣ Install dependencies

$ pip install -r requirements.txt

3️⃣ Setup & Preprocess Data

$ python preprocess.py --dataset BoT-IoT
$ python preprocess.py --dataset CICIDS2018

---

🏗 Model Architecture

1️⃣ Data Augmentation using cGAN

• Trains a conditional GAN to generate realistic synthetic minority class samples.
• Reduces class imbalance and enhances model generalization.

2️⃣ Dimensionality Reduction with Auxiliary Autoencoder

• Compresses high-dimensional network traffic data while preserving critical features.
• Achieves 93.02% data volume reduction on BoT-IoT and 96.25% on CICIDS2018.

3️⃣ Hybrid Transformer Model with Uncertainty Estimation

• Uses a Transformer-based model for traffic classification.
• Integrates Bayesian uncertainty estimation to improve reliability.

---

🚀 Training & Evaluation

1️⃣ Train the Model

$ python train.py --dataset BoT-IoT --epochs 100
$ python train.py --dataset CICIDS2018 --epochs 100

2️⃣ Evaluate the Model

$ python evaluate.py --dataset BoT-IoT
$ python evaluate.py --dataset CICIDS2018

---

📌 Citation

If you use this code in your research, please cite our IEEE IoT Journal paper:

@article{yourcitation2024,
  author    = {Your Name and Co-Authors},
  title     = {IoTSecUT: Uncertainty-Based Hybrid Deep Learning Approach for Superior IoT Security Amidst Evolving Cyber Threats},
  journal   = {IEEE Internet of Things Journal},
  year      = {2024}
}

---

🔗 Contact

For any inquiries, feel free to reach out: 📧 Email: [email protected] / [email protected] 💻 GitHub: Author