Uncertainty Quantification of Cyber Attacks on Connected Vehicles and Infrastructure Final Report


Multiple studies have explored different forms of connected vehicle applications, such as queue warning and cooperative adaptive cruise control (CACC), in standard wireless access in vehicular environments (WAVE), and dedicated short-range communication (DSRC) network environments.  A major focus of our ongoing research is to consider a hybrid vehicle-to-everything (V2X) infrastructure, one that supports multiple types of wireless networks. Our work has led to a system framework that allows WAVE applications to run in a system that is agnostic of the underlying network stack details. This research explores the uncertainty quantifications of cyber-attacks in V2X systems.   Our results are summarized as follows: (i) a single malicious on-board unit (OBU) can significantly impair the channel, which would result in a significant increase in the average data loss rate and communication latency; (ii) a CACC platoon can easily detect an unreliable data stream and can fall back gracefully to a variant of adaptive cruise control (ACC), which we refer to as eCACC (emulated CACC). eCACC uses a local smart sensor that can estimate the velocity and acceleration of the preceding vehicle (vehicle ahead) of a subject vehicle; (iii) if there is a noise associated with a DSRC on-board unit in a vehicle within the CACC platoon, the system must fall back to standard ACC; and (iv) local and global adaptation algorithms are designed to maximize traffic flow while ensuring platoon string stability. In the follow-up report of this project (Part 2), we will present two statistical models, specifically two change-point models, for real-time V2I cyber attack detection in a connected vehicle environment.