Adversarial Attacks for Intrusion Detection Based on Bus Traffic

Authors

Daojing He, Harbin Institute of Technology, Shenzhen, China & Guangdong Provincial Key Laboratory of Novel Security Intelligence Technologies
Jiayu Dai, East China Normal University, China
Xiaoxia Liu, East China Normal University, China
Shanshan Zhu, East China Normal University, China
Sammy Chan, City University of Hong Kong
Mohsen Guizani, Mohamed bin Zayed University of Artificial IntelligenceFollow

Document Type

Article

Publication Title

IEEE Network

Abstract

A communication bus is used to transmit electronic signals between components, realize functional integration through information sharing, and improve system efficiency. The current research on intrusion detection based on bus traffic is mainly pertaining to machine learning or time logic detection. However, recent studies have shown that machine learning models perform poorly in defense of various adversarial attacks. In this article, we propose a method based on generative adversarial networks to transform normal traffic into adversarial and malicious ones. To be closer to reality, adversarial example generation models on two threat scenarios are proposed. At the same time, the distance metric L2 is introduced in the loss function to ensure the authenticity of the generated adversarial examples. To evaluate our method, we use the traffic generated by the model to various intrusion detection systems based on bus. Experimental results show that the model is effective because the detection rate of different intrusion detection models decreases after the traffic is processed. Thus, the traffic generated by our models can be used as training data to enhance the accuracy of intrusion detection systems. IEEE

First Page

1

Last Page

7

DOI

10.1109/MNET.105.2100353

Publication Date

8-22-2022

Keywords

Data models, Generative adversarial networks, Intrusion detection, Ions, Protocols, Security, Training, Bayesian networks, Computer crime, Learning systems, Network security

Comments

IR Deposit conditions:

OA version (pathway a) Accepted version

No embargo

When accepted for publication, set statement to accompany deposit (see policy)

Must link to publisher version with DOI

Publisher copyright and source must be acknowledged

Recommended Citation

D. He, J. Dai, X. Liu, S. Zhu, S. Chan and M. Guizani, "Adversarial Attacks for Intrusion Detection Based on Bus Traffic," in IEEE Network,, 2022, doi: 10.1109/MNET.105.2100353.