Using Semantic Technologies to Mine Vehicular Context for Security

Abstract

The number of sensors, actuators and electronic control units present in cars have increased in the last few years. The Internet-of-Things (IoT) model has transformed modern vehicles into a co-engineered interacting network of physical and computational components. Vehicles have become a complex cyber-physical system where context detection has become a challenge. In this paper, we present a rule based approach for context detection in vehicles. We also discuss various attack surfaces and vulnerabilities in vehicular IoT. We propose a system which collects data from the CAN bus and uses it to generate SWRL rules. We then reason over these rules to mine vehicular context. We also showcase a few use-cases as examples where our system can detect if a vehicle is in an unsafe/anomalous state