Distributed Data-Driven Control for a Connected Autonomous Vehicle Platoon Subjected to False Data Injection Attacks

In this paper, we consider the need for deployment in the long-distance safe longitudinal formation control task when the connected autonomous vehicle (CAV) platoon is subjected to malicious cyber attacks. To ensure the safe, orderly, stable and efficient driving performance of the vehicle platoon, a novel distributed data-driven control (DDDC) approach for a homogeneous connected autonomous vehicle platoon under false data injection (FDI) attacks is investigated. First, an FDI attacks detection and compensation mechanism is designed to detect whether the received position signals are under attack or not and compensate the attacked position signals. Then, a novel DDDC approach for the vehicle platoon longitudinal formation control is developed by using the compensation data from the designed attack compensation mechanism and a dynamic linearization data model. Theoretical analysis verifies that the proposed DDDC method can ensure the internal stability (IS) and string stability (SS) of the homogeneous platoon subjected to FDI attacks. Finally, the effectiveness and practicality of the proposed DDDC approach are validated through a group of comparative simulations subjected to random FDI attacks of equal frequency and magnitude.
Note to Practitioners
—This work aims to solve the vehicle platoon long-distance safe longitudinal formation control task subjected to malicious FDI attacks. FDI attacks can achieve their destructive purposes by processing intercepted information and injecting false data into the original information. Existing literature overly relies on a priori knowledge of network attacks, yet in practice it is difficult to capture the true intentions of attackers in advance. For multi-channel V2V communication networks, it is even more important to design a resilient and accurate distributed controller strategy for such unpredictable and specific network attacks. Therefore, this paper proposes a data-driven distributed longitudinal formation control strategy with attack detection and compensation mechanism. The proposed strategy is shown to be able to ensure the safe longitudinal formation control task for the homogeneous CAV platoon suffering from FDI attacks. In addition, the stability of the CAV platoon is then investigated while the attacked signals are detected and cleaned, and it is shown to guarantee the internal stability of a single vehicle and the string stability of the platoon.