AuSSE: A Novel Framework for Security and Safety Evaluation for Autonomous Vehicles

Autonomous vehicles (AV) are becoming increasingly efficient and equipped with advanced technologies and connectivity. These include over-the-air software updates, connecting telematics data, and software-defined vehicles. However, these advancements also introduce new potential attack surfaces and increase AV security and safety risks. A critical challenge arises: How can we assess cyber attack impact on the operational safety of AV? To address this challenge, this paper proposes a novel framework named AuSSE (Autonomous Vehicle Security and Safety Evaluation).We developed a novel graphical security model called VHARM to identify attack paths and assess security risks. Additionally, we created CARLASec, an extension to the existing CARLA simulator. CARLASec enables the handling of in-vehicle networks, protocols, cyber attack injection mechanisms, and safety scores. As a result, it is suitable for a wider range of research on AV attack and defence. Additionally, recognising that AV security and safety cannot be separated, we introduce methods for comprehensive security and safety evaluations.