CAN-AE-Transformer-IDS

This is source code for paper "Multi-Classification In-Vehicle Intrusion Detection System using Packet- and Sequence-Level Characteristics from Time-Embedded Transformer with Autoencoder"

Journal: Knowledge-Based Systems

Link: https://doi.org/10.1016/j.knosys.2024.112091

Data Preprocessing

Car Hacking dataset

CUDA_VISIBLE_DEVICES=0 python3 splitDataIntoSession.py --window_size=29 --strided=29 --attack_type=chd > data_preprocessing_chd.txt

ROAD Fabrication dataset

CUDA_VISIBLE_DEVICES=0 python3 splitDataIntoSession.py --window_size=15 --strided=15 --attack_type=road_fab --indir=./road/fab_dataset --outdir=./road/fab_multi/TFRecord > data_preprocessing_roadfab.txt

ROAD Masquerade dataset

CUDA_VISIBLE_DEVICES=0 python3 splitDataIntoSession.py --window_size=15 --strided=15 --attack_type=road_mas --indir=./road/mas_dataset --outdir=./road/mas_multi/TFRecord > data_preprocessing_roadmas.txt

Train test split

Car hacking dataset

CUDA_VISIBLE_DEVICES=1 python3 trainTestSplit.py --data_path=./data/CHD --window_size 29 --strided 29 --rid 1

ROAD dataset

CUDA_VISIBLE_DEVICES=1 python3 trainTestSplit.py --data_path=./data/fab_dataset --window_size 15 --strided 15 --rid 1

CUDA_VISIBLE_DEVICES=1 python3 trainTestSplit.py --data_path=./data/mas_dataset --window_size 15 --strided 15 --rid 1

Train

ROAD masquerade dataset

CUDA_VISIBLE_DEVICES=2 python3 new_trainer.py --indir=./road/mas_multi/TFRecord_w15_s15/1/ --window_size=15 --batch_size=32 --type=road_mas --mode=cb --tse=True --epoch=150 --ver=2

ROAD fabrication dataset

CUDA_VISIBLE_DEVICES=2 python3 new_trainer.py --indir=./road/fab_multi/TFRecord_w15_s15/1/ --window_size=15 --batch_size=32 --type=road_fab --mode=cb --tse=True --epoch=300 --ver=2

Params

--indir, type=str | Data path input

--window_size, type=int | Window size (15 for ROAD and 29 for CHD)

--type, type=str | Dataset type: chd or road

--epoch, type=int | Epoch value

--batch_size, type=int | Batch size value

--lr, type=float | Learning rate

--mode, type=str | cb or ae (the ways of using AE)

--tse, type=bool | Using tse

--ver, type=str | Version value

Citation

If you use this code for your research, please cite our paper.

@article{LE2024112091,
title = {Multi-classification in-vehicle intrusion detection system using packet- and sequence-level characteristics from time-embedded transformer with autoencoder},
journal = {Knowledge-Based Systems},
volume = {299},
pages = {112091},
year = {2024},
issn = {0950-7051},
doi = {https://doi.org/10.1016/j.knosys.2024.112091},
url = {https://www.sciencedirect.com/science/article/pii/S0950705124007251}}