Anomaly Detection for In-Vehicle Networks using a Sensor-based Approach

Modern vehicles made a meaningful development to more complexity as well as connectivity in the last decade. Therefore, they cannot be seen as a closed system anymore. As the openness of the vehicle is increasing, so does the security risk for the in-vehicle networks and its components. Apart from threats for comfort and confidentiality, these attacks can also affect safety critical systems of the vehicle and hence endanger the driver and other road users. This paper discusses anomaly detection capabilities for in-vehicle networks based on sensors monitoring the internal network traffic. With respect to characteristics of typical vehicular networks, like the Controller Area Network (CAN), a recognition model for potential attacks during the operation of the vehicle without causing false positives is presented. Moreover, important design and application criteria for such an extension of the vehicle's security architecture are explained and discussed.