Data collected from a car's internal computer network, or its CAN bus, can identify its driver based on driving style in a study performed by researchers from the University of Washington and the University of California, San Diego.
By using the car's sensors collecting data on braking patterns, acceleration, and steering wheel angle, the team could distinguish the correct driver with 100% accuracy after 15 minutes of driving.
Fifteen test subjects each drove a short distance followed by a longer route, and trip data was recorded by a laptop connected to the car's CAN network. An algorithm then studied 90% of the driving data for patterns and applied that information to match a driver to the remaining data. It chose the correct driver based only on material gathered from the brake pedal within 15 minutes of driving 90% of the time.
The ability for a car to identify its driver could have unexpected privacy implications, the researchers note. For example, insurance companies and car rental agencies could fine customers for allowing unauthorized drivers behind the wheel, and former University of Washington research Miron Enev says CAN networks should have permission systems similar to the ones on smartphones. "You should approach every new application that you expose your data to on a need-to-know basis," he says.
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA
No entries found