Modern vehicles are highly automated and connected, which results in grater risk for remote abuse. The purpose of this thesis is to outline the information security threats in connected cars, and suggest penetration testing methodology tailored to connected cars.
Introduction explains how penetration testing of connected cars helps and improves its information security, and why it is advisable to use customized methodology. It is followed by an overview of connected car's information technology properties, that are combining two informational worlds. On the outside, the connected car communicates with the managed infrastructure, other cars and passengers through generally used protocols, such as Wi-Fi, GSM and Bluetooth. On the inside, dedicated ECU devices communicate using CAN protocol, which is specific and due to its history highly vulnerable. Then, comparison of information security properties in connected cars is made based on CIA triad, the core principles of information security. Further on, existing penetration testing methodologies are described with the intention to deduct which properties are usable with connected cars. Making threat models ensures that testing is consistent, efficient and repeatable. Finally, a specialized methodology is suggested. It differs from classical methodologies in the initial attack – the inside CAN network is examined separately through the OBD-II service port, since that option is available to the potential hacker as well. Suggested specialized methodology differs also in the lateral move, that in classical methodologies means cracking and catching passwords, and in connected car it has to do with shifting communication from Ethernet into the CAN protocol.
One of the major upcoming threats is introduced with mobile applications entering the infotainment system, which makes standardized penetration testing of connected cars a must.
|
