Integrating Cybersecurity Concerns in Functional Safety Assurance of AI-Based Automotive Functions

Automated vehicles are characterized by intelligent functions (many developed using AI-based solutions), autonomy, and high connectivity with the external environment. The high connectivity of AV opens relevant cybersecurity issues due to many potential attack surfaces. Cybersecurity attacks' effects on AV may also determine severe risks in terms of functional safety due to vehicles' high automation and autonomy. Functional safety assurance of AV cannot be disjoint from the analysis and treatment of cybersecurity concerns. In this paper, we define a methodology called ISCA (Integrated Safety and Cybersecurity Assurance) aimed at integrating, in a systematic manner, cybersecurity concerns into the functional safety assurance process of AI-based functions of AV. The ISCA methodology focuses on the emerging scenario-based approach for testing AI-based vehicle functions and is based on creating variants of the scenarios used for functional safety testing. The variants are derived from the original scenarios modified by the effects of highly risky cyberattacks that can occur during the scenario execution. The risk assessment of cyberattacks is based on the fuzzy logic mechanism. ISCA represents a possible way to strengthen the scenario-based functional safety assurance of AI-based vehicle functions by integrating cybersecurity concerns.