Visible to the public Vehicle Self-Surveillance: Sensor-Enabled Automatic Driver Recognition

TitleVehicle Self-Surveillance: Sensor-Enabled Automatic Driver Recognition
Publication TypeConference Paper
Year of Publication2016
AuthorsMarkwood, Ian D., Liu, Yao
Conference NameProceedings of the 11th ACM on Asia Conference on Computer and Communications Security
Conference LocationNew York, NY, USA
ISBN Number978-1-4503-4233-9
Keywordsauthentication, Behavioral biometrics, driving behavior modeling, Human Behavior, pubcrawl, Resiliency, Scalability, sensor security

Motor vehicles are widely used, quite valuable, and often targeted for theft. Preventive measures include car alarms, proximity control, and physical locks, which can be bypassed if the car is left unlocked, or if the thief obtains the keys. Reactive strategies like cameras, motion detectors, human patrolling, and GPS tracking can monitor a vehicle, but may not detect car thefts in a timely manner. We propose a fast automatic driver recognition system that identifies unauthorized drivers while overcoming the drawbacks of previous approaches. We factor drivers' trips into elemental driving events, from which we extract their driving preference features that cannot be exactly reproduced by a thief driving away in the stolen car. We performed real world evaluation using the driving data collected from 31 volunteers. Experiment results show we can distinguish the current driver as the owner with 97% accuracy, while preventing impersonation 91% of the time.

Citation Keymarkwood_vehicle_2016