A new kind of vulnerability
Many vehicles currently on Canadian roads feature Level-2 automation, meaning they can assist with steering, braking and lane changes. However, they still depend on the driver to remain attentive and ready to take over when prompted. “This dependency creates a novel class of vulnerabilities,” said Sarkar. “Modern cybersecurity frameworks focus on software and hardware. But they don’t adequately address risks rooted in driver behaviour, attention and trust in automation.” For example, if an attacker understands how drivers typically respond to alerts or how long it takes them to re-engage, that information could be exploited to maximize disruption or risk during a takeover request. Instead of treating the human as an external variable, Sarkar’s project integrates human behaviour directly into cybersecurity modelling. Her team is developing models and simulations that treat driver behaviour as part of the cybersecurity system. The goal is to design new safety protocols and defensive strategies that anticipate these risks before they manifest on public roads.
(L to R) PhD students Mayar Nour and Nour ElGharably test different driving scenarios using a simulator for semi-autonomous vehicles. (Jacob Arts/Western Engineering)












