Data-driven abnormal behavior detection for autonomous platoon

Abstract

Autonomous platoon is a technique where co-operative adaptive cruise control (CACC) enabled vehicles are organized into groups of close following vehicles through communication. It is envisioned that with the increased demand for autonomous vehicles, platoons would be a part of our life in near future. Although many efforts have been devoted to implement the vehicle platooning, ensuring the security remains challenging. Due to lack of security, platoons would be subject to modified packets which can mislead the platoon and result in platoon instability. Therefore, identifying malicious vehicles has become an important requirement. In this paper, we investigate a data-driven abnormal behavior detection approach for an autonomous platoon. We propose a novel statistical learning based technique to detect data anomalies. We demonstrate that shared speed value among platoon members would be sufficient to detect the misbehaving vehicles.