CAN intrusion dataset for IDS development and digital twin simulation

In-vehicle networks of modern automobiles, particularly Controller Area Networks (CAN), are vulnerable to cyberattacks which can cause dangerous accidents. There are many CAN intrusion detection systems (IDS) that use different techniques and features. While there are several CAN datasets available from real vehicles and CAN testbeds, we identify a lack of datasets containing signals decoded from raw CAN messages, which hinders the development and evaluation of signal-based CAN IDS and prevents their comparison with other types of CAN IDS. We aim to address this gap by using open-source CAN databases to create a decoded signal dataset from raw CAN data and identify the physical values represented by the signals. Such a dataset can be used to create digital twin simulations that allow the evaluation of CAN IDS in a realistic and safe test environment and the generation of advanced attack scenarios that are lacking in current datasets. This collaborative project shall facilitate the sharing of knowledge, expertise, and insights into industry trends and best practices, as well as resources such as specialised tools and facilities, between the participating institutions. It will also enable participating researchers to expand their networks, leading to further opportunities and contributions to the area of cybersecurity.

Faculty Supervisor:

Arash Habibi Lashkari

Student:

Partner:

International Islamic University Malaysia

Discipline:

Computer science

Sector:

Cyber Security; Automotive; Information and Communications Technology

University:

York University

Program: