The role of Machine Learning in Vehicular Cyber Security
By David Ngetich
There is no doubt that cybersecurity is the front-end-and-center topic in global conversation pertaining policy, research and development of new technology. There is a vital need in ensuring that the consumers of the tech products are safer from the malicious intent and cyber harm. As it concerns self-driving cars the car-makers need to prioritize the safety of the passengers. In doing so, a lot of car-makers are turning to machine learning as means to help them bolster vehicular cybersecurity.
In August 2018, during the Global Automotive Cybersecurity Summit, the cybersecurity of the autonomous and connected vehicles was of key concerns to all involved. The Summit did identify the use of machine learning to fight cyber-attacks on self-driving cars.
According to Dino Causevic, a Cybersecurity Engineer at Toptal, autonomous vehicles can employ Machine Learning algorithms to help fight the cyber-attacks targeted at self-driving vehicles. He suggests that the first phase of deploying AI to fight this security problems would be by collecting and storing the right data of the car’s instructions. Once the right data has been stored, the algorithm will run these data against any other instruction that the car is given while driving. In essence, the algorithm has learned the normal behavior and can flag off any anomalies.
For instance, developing in the car’s internal network system capable of storing and analyzing logs hence giving the car the capacity to prevent attacks immediately it detects an anomaly. These implemented machine learning algorithms can be used to detect malware activities, unusual commands such as activating the parking mode while the car is driving at 80 mph on the highway.
These “learn and prevent” mechanism has already been implemented by Miller and Valasek who were able to hack into the Jeep Cherokee in 2015. They were able to showcase the vulnerability of the self-driving cars but also came with a solution that uses Machine Learning to detect and prevent/report these cyber-attacks.
