GateKeeper

The repository of GateKeeper, code for our Computer Networks Journal paper: GateKeeper: An UltraLite malicious traffic identification method with dual-aspect optimization strategies on IoT gateways

Note:

• ⭐ Please leave a STAR if you like this work! ⭐

Dependencies

Ensure the following dependencies are installed:

• Python 3.7+
• PyTorch 1.4.0+

Dataset

• ToN-IoT
• IoT-23
• CIC-IoT-2022

File Structure

• run_Base.py: Script to run the Base model.
• train.py: Contains functions for training and testing the model.
• Base.py: Defines the structure of the Base model.
• GateKeeper.py: Defines the structure of the GateKeeper model.
• KBS_score.py: Script for calculating and evaluating attention scores.
• utils_base.py: Utility functions for dataset construction and iterators.
• utils_GateKeeper.py: Utility functions specific to the GateKeeper model.

Usage

Adjust ./dataset to your data

(1) Train the Base Model

python run_Base.py --test False

(2) Get the importance score of each byte

python KBS_score.py

(3) Train the GateKeeper

Copy the results of importance score (KBS_score.py print) and assign them to the pos variable in utils_GateKeeper.py.

python run_GateKeeper.py --test False

Please quote if it helps you

@article{cao2024gatekeeper,
  title={GateKeeper: An UltraLite malicious traffic identification method with dual-aspect optimization strategies on IoT gateways},
  author={Cao, Jie and Xu, Yuwei and Yu, Enze and Xiang, Qiao and Song, Kehui and He, Liang and Cheng, Guang},
  journal={Computer Networks},
  pages={110556},
  year={2024},
  publisher={Elsevier}
}
@inproceedings{cao2023mathcal,
  title={$$\backslash$mathcal $\{$L$\}$$\{$-$\}$ $ ETC: A Lightweight Model Based on Key Bytes Selection for Encrypted Traffic Classification},
  author={Cao, Jie and Xu, Yuwei and Xiang, Qiao},
  booktitle={ICC 2023-IEEE International Conference on Communications},
  pages={2370--2375},
  year={2023},
  organization={IEEE}
}