Scaling a Real-Time Driving Behaviour Analysis System

Drivian is a system that helps drivers save fuel and drive safer. The Drivian system gathers data from smartphone sensors, analysis it, and provides real-time feedback to the user on his driving style, in terms of security and economy. Also, Drivian alerts the driver for harsh acceleration, braking and cornering events. Moreover, the collected data is forwarded to a central location and used for tracking the driver evolution over time, record trips, extract statistical information, validate trips and develop new insights. Considering that each device produces a new tuple of sensor data every 60ms, with several thousand (or more) equipments online, the amount data to process and store in real-time presents a serious challenge. In this talk, we will portray and discuss the major design decisions for coping with the architectural drivers for the Drivian system.