Detecting subtle cyberattacks on adaptive cruise control vehicles: A machine learning approach

TL;DR

This work addresses the detection of subtle cyberattacks on ACC-equipped vehicles by modeling three attack classes within car-following dynamics, and developing a GAN-based trajectory anomaly detector for real-time identification. The approach learns the distribution of normal vehicle trajectories using a 1D-CNN GAN and detects attacks by reconstructing observed data in latent space and evaluating a composite loss. Numerical results show that attacks on acceleration (Type i@) more strongly disrupt micro- and macro-scale traffic, while false-data injection (Type ii@) and DoS (Type iii@) have milder or different perturbation effects; the detector achieves high recall and competitive precision and outperforms several baselines. The findings underscore the potential network-wide consequences of even a subset of compromised ACC vehicles and demonstrate a promising, real-time defense mechanism with implications for resilient traffic management and automated-vehicle security.

Abstract

With the advent of vehicles equipped with advanced driver-assistance systems, such as adaptive cruise control (ACC) and other automated driving features, the potential for cyberattacks on these automated vehicles (AVs) has emerged. While overt attacks that force vehicles to collide may be easily identified, more insidious attacks, which only slightly alter driving behavior, can result in network-wide increases in congestion, fuel consumption, and even crash risk without being easily detected. To address the detection of such attacks, we first present a traffic model framework for three types of potential cyberattacks: malicious manipulation of vehicle control commands, false data injection attacks on sensor measurements, and denial-of-service (DoS) attacks. We then investigate the impacts of these attacks at both the individual vehicle (micro) and traffic flow (macro) levels. A novel generative adversarial network (GAN)-based anomaly detection model is proposed for real-time identification of such attacks using vehicle trajectory data. We provide numerical evidence {to demonstrate} the efficacy of our machine learning approach in detecting cyberattacks on ACC-equipped vehicles. The proposed method is compared against some recently proposed neural network models and observed to have higher accuracy in identifying anomalous driving behaviors of ACC vehicles.

Figures (8)

• Figure 1: Overview of the proposed GAN-based attack detection strategy where typical trajectory data is used to construct a GAN for generating synthetic trajectory data, which is then applied to newly observed trajectory data to identify anomalous traffic flow.
• Figure 2: Demonstration of a GAN-based attack detection algorithm to measure detection score.
• Figure 3: Trajectories of speed and position without attack for human-driven vehicles, in green; and ACC vehicles, in blue.
• Figure 4: Trajectories of speed and position for ACC vehicles under Type i@ attack, in red; human-driven vehicles, in green; and unattacked ACC vehicles, in blue.
• Figure 5: Trajectories of speed and position for ACC vehicles under Type ii@ attack, in red; human-driven vehicles, in green; and unattacked ACC vehicles, in blue.