README.md

FOSS: Towards Fine-Grained Unknown Class Detection against the Open-Set Attack Spectrum with Variable Legitimate Traffic

Introduction

FOSS is the anomaly-based network intrusion detection system which aims to achieve: (i) fine-grained unknown attack detection and (ii) ever-changing legitimate traffic adaptation. The architecture of FOSS mainly includes model construction, outlier detection & classification, and model update. This anonymous repository displays the corresponding source code for model implementation.

Requirements

pip install scipy
pip install numpy
pip install pandas
pip install tqdm
pip install pyecharts
pip install joblib
pip install pickle
pip install sklearn

Feature Extraction

The feature extraction program is stored in ./overview/feature.py. Please configure the data input path that includes PCAP traffic split by 5-tuple.

python feature.py

Model

The main function is stored in ./model/main_process.py.

python main_process.py

Evaluation

The folder ./evaluation/monte_carlo/ shows the feature selection based on the weighted entropy in the Monte Carlo method.

The feature perception evaluation results for 8 types of attacks from IDS are shown in ./evaluation/feature_perception/.

References

• Kitsune: An Ensemble of Autoencoders for Online Network Intrusion Detection, Yisroel Mirsky, Tomer Doitshman, Yuval Elovici, and Asaf Shabtai - NDSS 2018
• Realtime Robust Malicious Traffic Detection via Frequency Domain Analysis, Chuanpu Fu, Qi Li, Meng Shen, and Ke Xu - CCS 2021
• FARE: A Flow Sequence Network For Encrypted Traffic Classification, Junjie Liang, Wenbo Guo, Tongbo Luo, Vasant G. Honavar, Gang Wang, and Xinyu Xing - NDSS 2021
• Random Partitioning Forest for Point-Wise and Collective Anomaly Detection - Application to Network Intrusion Detection, Pierre-Francois Marteau - TIFS 2021
• Classification Under Streaming Emerging New Classes: A Solution Using Completely-Random Trees, Xin Mu, Kai Ming Ting, and Zhi-Hua Zhou - TKDE 2017
• Conditional Variational Auto-Encoder and Extreme Value Theory Aided Two-Stage Learning Approach for Intelligent Fine-Grained Known/Unknown Intrusion Detection, Jian Yang, Xiang Chen, Shuangwu Chen, Xiaofeng Jiang, and Xiaobin Tan - TIFS 2021