ETD

• Automotive
• Autosar
• Embedded
• IDS
• Sybersecurity

The increasing complexity of modern vehicles, driven by the proliferation of electronic control units and advanced functionalities, has made in-vehicle com- munication networks particularly the Controller Area Network (CAN) bus a fundamental component of automotive architectures. As vehicles have be- come more connected through external interfaces, wireless communications, and over-the-air updates the assumption of secure communication on CAN is no longer valid. The CAN protocol was designed to be simple and efficient, which leads to security vulnerabilities, as it does not implement methods dedicated to ensuring communication security. Intrusion Detection Systems (IDSs) represent a promising approach to address these challenges. An auto- motive IDS monitors network traffic and system behavior to identify anoma- lies or attacks by comparing observed patterns against learned models of nor- mal operation. Depending on the adopted approach, it can rely on rule-based or machine learning techniques, cryptographic methods for secure commu- nication and message validation, and physical-layer analysis such as voltage levels and timing features to enhance detection capabilities and identify so- phisticated intrusions. The objective of this thesis is to develop a framework for the inclusion of IDSs based on different features into a multi-feature IDS that is hardware-independent and compliant with the AUTOSAR standard for IDS in the automotive sector. The framework offers a layered development environment that allows integration of non-AUTOSAR compliant IDSs that perform anomaly detection based on various parameters such as message fre- quency, voltage difference recorded upon message arrival, and cryptographic challenge-response timing. During the development and testing phase, a test bench was created with boards to simulate the multi-features IDS and various ECUs, recreating a real environment with constant traffic. The testing phase involved validating the correct integration of figerprinting based IDSs that had already been developed, as well as validating the execution flow of the framework to ensure its compatibility with the AUTOSAR standard.