AdvBiTrans: A Hybrid BiLSTM-Transformer with Adversarial Training for Botnet Detection in IIoT

This repository contains the complete implementation of our paper "AdvBiTrans", a robust and accurate deep learning framework for botnet detection in Industrial Internet of Things (IIoT) environments.

---

📂 Project Structure

AdvBiTrans_experiment/ │ ├── CIC-DDoS2019/ │ ├── 1.Data_Preprocessing.ipynb │ ├── 2.Feature_Selection_PCCS.ipynb │ ├── 3.Mult_Model_xx.ipynb │ └── Picture/ │ ├── N-BaIoT/ │ ├── 1.Data_Resaving.ipynb │ ├── 2.Data_Preprocessing.ipynb │ ├── 3.Feature_Selection_PCCS.ipynb │ ├── 4.Mult_Model_xx.ipynb │ └── Picture/ │ ├── requirements.txt └── README.md

📋 Description

The project includes:

• 📊 Data preprocessing for CIC-DDoS2019 and N-BaIoT datasets
• 🧪 Feature selection using Pearson Correlation Coefficient Strategy (PCCS)
• 🤖 Model training and evaluation for AdvBiTrans, 1D-CNN, MLP, LightGBM, TabNet, and XGBoost
• 🔐 Adversarial training using PGD (Projected Gradient Descent)
• 📈 Performance metrics including Accuracy, F1, Inference Time, Memory Usage, FLOPs Note: Mult_Model_xx is named according to the number of features to train the model, the model and parameters are consistent.

---

🛠️ Requirements

Run the following to install the required Python packages:

pip install -r requirements.txt


## 📊 Datasets
The foundational datasets utilized in this paper are publicly accessible. The N-BaIoT dataset can be obtained from the official repository at
http://archive.ics.uci.edu/ml/datasets/detection_of_IoT_botnet_attacks_N_BaIoT.
The CIC-DDoS2019 dataset is available at https://www.unb.ca/cic/datasets/ddos-2019.html.

## 🧪 Run Experiment
You can open and execute each .ipynb notebook in order using Jupyter Notebook or VSCode with Python extension.