User Profiling of Automobile Driver and Outlier Detection

The central idea of the paper is to profile the driver of vehicle. Profiling is the collection of data associated with specific user. It’s the representation of person's identity. So we will profile and learn the driver behavior from his driving habits like the way he change gears, the way he accelerates, the way he takes turn and many more parameters such that any unusual change in driving behavior can be traced and should be reported to the authentic user in real time. All the data will be recorded in the vehicle itself using in vehicle data recorder (IVDR). There are certain diagnostic codes made available by auto manufacturer at the OBD-II port, Controller Area Network (CAN) data recorder of the vehicle. All this learning process is carried out implicitly; no explicit involvement of driver is required other than driving the vehicle. Our focus is on household vehicle generally being driven by various members of family and that’s make the vehicle as "Multiuser Single System".