• automotive cybersecurity
• electronic control units
• intrusion detection systems
• runtime fingerprinting

The last decade has seen an increase in the functionalities present in vehicles, from autonomous driving to infotainment, diagnostics, and extended connectivity; these advances have led manufacturers to include an ever-increasing number of Electronic Control Units (ECUs) interconnected to implement increasingly sophisticated tasks. The increase in devices and the widening of the attack surface has opened the door to intrusions and attacks in vehicles, with consequences that can be different depending on the attacked domain. Over the years, countermeasures have been developed to overcome the lack of encryption and authentication of the CAN-bus; and in-vehicle Intrusion Detection Systems (IDSs) have increasingly established themselves as a solution. These systems are mainly AI-based, rule-based, or hardware fingerprinting-based. The main objective of this thesis is to explore the possibility of runtime fingerprinting of ECUs through the execution of predefined software tasks. Cryptographic tasks for ECUs were developed to ensure their authentication, and machine learning models were used to classify the execution times of tasks by ECUs. To replicate the architecture of a vehicle, a specific experimental setup was created, a development pipeline was designed, and the performance of the developed IDS was analyzed.