Contributions to in-vehicle networks: error injection and intrusion detection system for CAN, and audio video bridging synchronization

Abstract

Vehicles are equipped with Electronic Control Units (ECUs) responsible for performing tasks as simple as lighting up brake lights or as complex as controlling the wheels of an electric car. The exchange of information between ECUs uses the Controller Area Network (CAN) and the Controller Area Network with Flexible Data Rate (CAN FD), which provides higher data rates and payloads, as the main and most used intra-vehicular networks, at least until today. Interconnected ECUs must work perfectly and interact with each other as well as with other car components in a reliable way, thus it is indispensable to test and predict the behavior of these units in error situations. For this, an error injection mechanism can be very advantageous for checking various error conditions in real-world scenarios that affect the safety of vehicles. Furthermore, nowadays, cars are also equipped with network technologies that provide connectivity to the external world. This offers numerous possibilities in terms of new applications and services to be provided, however makes the car a network node subject to cyber-attacks. It is then necessary to provide security mechanisms to prevent, or at least detect, attacks. Besides CAN and CAN FD networks, the advent of 100BASE-T1 Ethernet has feasible the outcome of many new automotive applications with higher bandwidth demands. In order to be use Ethernet in applications that require determinism, a series of IEEE standards, which together constitute the Audio Video Bridging (AVB), were developed and proposed. The IEEE 802.1AS, for instance, is the AVB standard that defines the generalized Precision Time Protocol (gPTP), responsible for node synchronization within AVB networks. In this context, this dissertation intends to make contributions for CAN/CAN FD networks as well as for the automotive Ethernet. For CAN networks, it proposes a novel Error Injection Technique to assist with system level validation tests and also an Intrusion Detection System based on machine learning algorithms. For automotive Ethernet, it designs and proposes a hardware implementation of the gPTP protocol that achieves the required nanoseconds precision, while also providing implementation details for future researches on that protocol.