Linxi Zhang
I am Linxi Zhang, a tenure-track assistant professor in the Department of Computer Science at Central Michigan University.
I obtained my Ph.D. degree in Computer and Information Science at University of Michigan Rackham Graduate School
under the guidance of Dr. Di Ma.
I am broadly interested in automotive cybersecurity, intrusion detection systems, machine learning, and wireless network and mobile system security.
Currently, I am focusing on in-vehicle network security and machine learning-based intrusion detection systems for CAN bus.
I am looking for self-motivated students. Visiting students and scholars are all welcome.
If you are interested in my research and working with me, please contact me directly with your CV.
Research
Journal Papers
- A Binarized Neural Network Approach to Accelerate In-Vehicle Network Intrusion Detection
Linxi Zhang, Xuke Yan, Di Ma
IEEE Access 2022 | Link - Accelerating In-Vehicle Network Intrusion Detection System using Binarized Neural Network
Linxi Zhang, Xuke Yan, Di Ma
SAE International Journal of Advances and Current Practices in Mobility 2022 | [Accepted] - A Hybrid Approach toward Efficient and Accurate Intrusion Detection for In-Vehicle Networks
Linxi Zhang, Di Ma
IEEE Access 2022 | Link
Conference Papers
- A Lightweight and Fast Approach for Upper Limb Range of Motion Assessment
Xuke Yan, Linxi Zhang, Bo Liu, Guangzhi Qu
ICMLA 2022 | [Accepted] - Accelerating In-Vehicle Network Intrusion Detection System using Binarized Neural Network (Best Paper Award)
Linxi Zhang, Xuke Yan, Di Ma
SAE WCX 2022 | Link - A Two-Stage Deep Learning Approach for CAN Intrusion Detection
Linxi Zhang, Lyndon Shi, Lyndon Kaja, Di Ma
GVSETS 2018 | Link
Posters and Others
- A Lightweight and Fast Approach for Upper Limb Range of Motion Assessment
Xuke Yan, Linxi Zhang, Bo Liu, Guangzhi Qu
OU SECS Faculty Research Expo 2022 - A Hybrid Approach toward Efficient and Accurate Intrusion Detection for In-Vehicle Networks (Honorable Mention - Graduate Award)
Linxi Zhang, Di Ma
WiCyS 2022 - Accelerating In-Vehicle Network Intrusion Detection System using Binarized Neural Network
Linxi Zhang, Xuke Yan, Di Ma
UMD CECS Doctoral Poster Showcase 2022 - A Two-Stage Deep Learning Approach for CAN Intrusion Detection
Linxi Zhang, Di MA
UMD CECS Research Poster Session 2019 - Controller Area Network Intrusion Detection and Response System
Zachary Cavazos, Abigail Darmofal, Colton Holoday, Cody Liu, Lee Moore, Alisha Patel, Lyndon Shi, Linxi Zhang, Di Ma
WiCyS 2018
Teaching
Courses Taught at UM-Dearborn
Guest Lecture Given at UM-Dearborn
Projects
Workflow of the proposed IDS, in which we design two main components: an input generator and a binarized neural network model. The binarized neural network is a very lightweight network, which is suitable for low-capacity devices like vehicle ECU. We had a demonstration that the proposed BNN model can be deployed in an FPGA device for further processing acceleration.
Anomaly-based Intrusion Detection System (IDS) is considered an effective approach to secure CAN and detect malicious attacks.
Rule-based approach is efficient but limited in the detection accuracy while machine learning-based detection has comparably higher detection accuracy but higher computation cost at the same time.
In this paper, we propose a novel hybrid IDS that combines the benefits of both rule-based and machine learning-based approaches.
More specifically, we use machine learning methods to achieve a high detection rate while keeping the low computational requirement by offsetting the detection with a rule-based component.
This IDS is supposed to be placed on a central gateway or installed on an ECU that can monitor the whole CAN bus traffic.
Our experiments with CAN traces collected from four different vehicle models demonstrate the effectiveness and efficiency of the proposed hybrid IDS.
