Locality sensitive hashing for tampering detection in automotive systems
In modern auto vehicles we find dozens of Electronic Control Units (ECUs) running several hundred MBs of code, alongside sophisticated dashboards with integrated wireless communications. While this technological advancement has brought upon a wide range of advantages and integrated features, it also exposed the modern vehicle to significant cyber threats, as documented in prior works. Unfortunately, besides traditional cyber attacks, the security and normal operation of the modern vehicle are nowadays exposed to a different kind of threat. This is the tampering, which denotes a procedure that alters the vehicle's behavior in order to gain particular advantages (e.g., financial, operational). A fundamental distinction between tampering and cyber attacks, is that tampering occurs with the owner's consent. This paper presents an approach for detecting tampering within modern vehicles by leveraging the advantages of sensitive hashing, namely the Exact Euclidean Locality Sensitive Hashing (E2LSH) method. Experimental results based on a dataset collected from the On-Board Diagnostics port (OBD) of a Kia SOUL vehicle demonstrate the practical applicability of the developed methodology.