Cooperative Localization of Connected Vehicles With Attack Detection and Secure Fusion

The vulnerability from cyberattack undermines the security in cooperative localization of connected vehicles. This paper presents a localization framework for connected vehicles subject to false data injection attacks, featuring an attack detection scheme to get rid of abnormal position information from malicious vehicles. To this end, a refined kinematic model, measurement model, and attack model are introduced. By introducing a local Kalman decomposition procedure to extract the position measurements from the collection of connected vehicles, a fusion algorithm is presented to guarantee secure and accurate vehicle localization. The simulation and experimental results show that the method outperforms the state-of-the-art vehicle localization methods in urban environments.