Selective Sensing for Transportation Safety Jonathan Voris
N. Sertac Artan
Wenjia Li
[email protected]
[email protected]
[email protected]
Computer Science Department
Electrical & Computer Engineering Department
Computer Science Department
New York Institute of Technology
Driver Data Collection • Amount of driver data being recorded is increasing • Many new devices and applications
2
Potential Privacy Issues • Devices may record a variety of sensitive information including: • • • •
Geolocation Audio Images Instantaneous engine readings
3
Potential Security Issues • Modern cars controlled by Electronic Control Units (ECUs) connected by a Controller Area Network (CAN bus)
Sensors and communication interfaces available on a modern vehicle.
Examples of CAN Connections [1] 4
Potential Security Issues • Devices connect to a vehicle’s CAN bus via an on-board diagnostics (OBD)-II port
• Increases attack surface of critical components • Many devices also feature a wireless uplink
5
Driver Modeling Dilemma • Research challenge: how to enable emerging driving applications while ensuring • •
Driver privacy Vehicular security
6
Solution Idea: Selective Driver Data Modeling • Decouple sensing from critical vehicle systems • Measure involuntary driving habits to discern driver identity • Potential modalities: Steering behavior • Speed control characteristics • Indicator usage • Contextual road features •
7
Advantages of Selective Driver Data Modeling • Driver identity verification would eliminate fraud
• Deviations from past driving patterns can detect safety issues • Would not require direct access to a vehicle’s CAN bus
8
Related Work • Recent work has demonstrated the feasibility of vehicle network compromise [4][5]
• Pseudonym-based approaches to driving privacy have been proposed [3] • Previous work on modeling driver behavior have been limited in scale [2] • Some involve intrusive sensors [6]
9
Previous Results: Mobile Device Usage Modeling • Approach has been applied to identifying mobile device users • 10 minute time-to-detection with a 1 false positive per day
Mobile Device Usage Model ROC Curve
10
Current Research • Analyzing features for viability as driver identifiers • Assessing driving simulators for human subject study
11
Upcoming User Study • Ask to complete a simulated driving task • Provide different scenario descriptions: Typical • Leisurely • Rushed • Safe • Unsafe •
• Record driving characteristics via software •
And hardware, where possible
• Attempt to identify driver classes via selective data modeling 12
Applications to Vehicular Networks • Major research challenges in vehicular networks: • • •
VANET Infrastructure
Unreliable and error-prone communication media Constrained node energy supply Sensors in vehicular networks can be compromised
• Data is important to transportation system safety and efficiency • Modeling techniques needed to distinguish trustworthy data from untrustworthy data 13
Future Plans • Driving feature analysis • Development of vehicle-independent sensing hardware • If simulation study is promising, performing follow-up with real vehicles
14
Conclusion • Novel applications offer benefits to drivers • Must be wary of potential security and privacy issues • Careful applications of sensing can potentially collect required data while providing security • Potential applications to: • •
Fraud detection Vehicular networks
15
Thank you!
16
References [1] Fortin Electronic Systems. “What is CAN Bus?” Available at: http://canbuskit.com/what.php, 2015. [2] M. Khalid, A. Wahab, and N. Kamaruddin. "Real time driving data collection and driver verification using CMAC-MFCC," International Conference on Artificial Intelligence, 2008. [3] R. Lu, X. Li, T. Luan, X. Liang, and X. Shen, "Pseudonym changing at social spots: An effective strategy for location privacy in VANETs." Transactions on Vehicular Technology, 2012. [4] C. Miller, C. Valasek. "Adventures in automotive networks and control units," DEFCON 21, 2013. [5] C. Miller, C. Valasek. "Remote Exploitation of an Unaltered Passenger Vehicle," DEFCON 23, 2015. [6] A. Picot, S. Charbonnier, and A. Caplier, “On-Line Detection of Drowsiness Using Brain and Visual Information,” IEEE Transactions on Systems, Man, and Cybernetics, 2012. 17