AI-Enabled Hardware-in-the-Loop Validation for Automotive Cybersecurity: A Review of Cyber Threats, Testbeds, and Intelligent Detection

Abstract

Cybersecurity has become one of the most critical challenges in the intelligent and connected vehicle ecosystem of today. As modern vehicles become increasingly connected and intelligent, the frequency and sophistication of cyberattacks targeting automotive systems continue to grow at an alarming rate. Ensuring robust detection and prevention mechanisms has therefore become essential to safeguard driver safety and vehicle integrity. Rapid and accurate identification of cyberthreats is critical, as such attacks can disrupt vital Electronic Control Units (ECUs) and compromise functions such as braking, steering, or communication networks. This review provides a comprehensive analysis of the major categories of cyberattacks targeting modern vehicles, including physical, remote, in-network, firmware- and software-based, cloud- and connectivity-related, and sensor-level perception attacks. Contemporary vehicle architectures, connected-vehicle technologies, software-update mechanisms, and current automotive cybersecurity standards and regulations are also considered. Although traditional cybersecurity testing approaches offer valuable insight into software vulnerabilities, they fail to capture the full cyber–physical interactions that govern vehicle behavior under malicious conditions. In this review, we highlight the essential role of Hardware-in-the-Loop (HIL) and Vehicle-in-the-Loop (VIL) platforms as realistic and safe environments for evaluating the impact of cyberattacks on automotive control systems and for generating synchronized cyber–physical data under controlled attack scenarios. We further examine how artificial intelligence (AI) techniques contribute to detecting, mitigating, and countering these cyberthreats, including supervised and unsupervised intrusion detection, deep-learning-based temporal modeling, cyber–physical anomaly detection, and the emerging challenge of adversarial attacks against AI-based detectors. By synthesizing insights from automotive cybersecurity, HIL-/VIL-based validation, automotive cybersecurity datasets, and AI-driven intrusion detection, this paper establishes a foundation for developing and evaluating more resilient and secure connected and software-defined vehicle architectures.