ML-Based CAN Bus Message Sniffing and Classification

This paper presents the development of an advanced Controller Area Network (CAN) bus analyzer system, incorporating Machine Learning (ML) techniques to elevate cybersecurity in vehicle networks and diagnostics. The proposed system is designed to efficiently monitor, record, and analyze CAN traffic, providing critical insights into vehicle function and security vulnerabilities. Leveraging sophisticated ML algorithms, the system deciphers complex patterns within CAN signals, enabling capabilities such as secure command injection and reverse engineering of vehicle functions, which is a key asset for cybersecurity professionals in threat assessment. Notable innovations include automatic reverse engineering, achieved through linear regression analysis of action-induced sequences, and dynamic state prediction, enabled by Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRU). This approach enhances both the resilience and functionality of automotive cybersecurity measures, contributing a powerful tool to the field of connected vehicle security.