Over the years, cars have become increasingly complex electronic systems.
These electronic systems, grouped together in control units (ECUs) to reduce the length of electrical interconnections, communicate with each other via communication channels that are not protected against cyber attacks.
Moreover, interfacing with the outside world (Wi-Fi, Bluetooth, USB, 5G, ...) increases the attack surfaces.
It has been demonstrated how a vulnerability has made it possible to take full remote control of a car in order to stop it or distract the driver, putting him at risk of an accident.
Several methods of attack detection have been proposed in the literature, but all fail to recognise the source of the attack, i.e. the attacker.
In this thesis a fingerprinting technique is proposed, developed and implemented on a real-time embedded system on the basis of an algorithm that makes use of a purely statistical approach to univocally identify the various ECUs on the basis of the different behaviour they have at a physical electrical level (voltage-based).