Automotive safety integration using drunk driving behavior detection and multi-factor authentication -

Abstract

With the emergence of the self-driving cars, and the expansion of the electronic controls of the modern vehicle, there exist a need for better security measures and self-policing capabilities. Considering the emerging use of 3rd party wireless connected car dongles, we explored building a prototype to expand upon the existing solution’s safety and security features. First, we investigated several approaches to connect our prototype to the car’s CAN Bus interface. Using messages collected from the car’s OBD-II port, it detects the signals: Ignition, location, trip start, trip end, and car shutdown. It also collects several can bus signals like speed, rpm, engine load, throttle and other OBD-II PIDs (Parameter IDs) . Using these signals and telematics, we can have second factor authentication-notification systems for the important events. These events may include unauthorized car start, unauthorized route, abnormal driving behavior like drunk and intoxicated driving, out of bounds, and any other odd behavior. The acknowledgement or decline of these events may be handled differently according to severity. For example, an un-authorized ignition may not allow the car to start, while a minor intoxicated behavior detection may send a notification to the loved ones, and a major detection of an intoxicated driver may lead the car to gradually stop and notify the authorities. To achieve this, the prototype utilizes an existing solution, ‘Acceptto’, that offers second-factor authentication capabilities. For the rest of the thesis, we propose a method to detect drunk driving patterns using only basic car sensors, available through off the shelf OBD-II dongles. The sensor data include standard On-Board Diagnostic sensor information along with an accelerometer sensor and GPS coordinates which are provided by the dongle. We collect the information through drive tests of normal driving behavior and controlled drunk driving behavior. The controlled driving emulation reflects the most common cues