MIDS: Detecting Stealthy Masquerade and Tampering Attacks on CAN Bus via Bidirectional Mamba
Researchers have developed a new intrusion detection system called Mamba Intrusion Detection System (MIDS) specifically designed to combat stealthy masquerade and tampering attacks on a vehicle's Controller Area Network (CAN) bus. Unlike existing systems that focus on simpler attacks, MIDS utilizes a bidirectional selective state-space model to analyze CAN identifiers and payloads in parallel, reconstructing their joint temporal semantics. Tested on a physical Tesla Model 3 and four public benchmarks, MIDS achieved high F1 scores, outperforming existing methods by a significant margin and demonstrating its capability for real-time onboard deployment. AI
IMPACT Enhances automotive cybersecurity by providing a more robust defense against sophisticated internal threats on vehicle networks.