A Deep Learning Framework for Driving Behavior Identification on In-Vehicle CAN-BUS Sensor Data

Abstract

Human driving behaviors are personalized and unique, and the automobile fingerprint of drivers could be helpful to automatically identify different driving behaviors and further be applied in fields such as auto-theft systems. Current research suggests that in-vehicle Controller Area Network-BUS (CAN-BUS) data can be used as an effective representation of driving behavior for recognizing different drivers. However, it is difficult to capture complex temporal features of driving behaviors in traditional methods. This paper proposes an end-to-end deep learning framework by fusing convolutional neural networks and recurrent neural networks with an attention mechanism, which is more suitable for time series CAN-BUS sensor data. The proposed method can automatically learn features of driving behaviors and model temporal features without professional knowledge in features modeling. Moreover, the method can capture salient structure features of high-dimensional sensor data and explore the correlations among multi-sensor data for rich feature representations of driving behaviors. Experimental results show that the proposed framework performs well in the real world driving behavior identification task, outperforming the state-of-the-art methods.

Lightweight Driver Behavior Identification Model with Sparse Learning on In-Vehicle CAN-BUS Sensor Data

“…Driving information can be captured using several different sources, which mainly include in-vehicle sensors and smartphone sensors. In References [ 5 , 6 , 7 , 8 , 10 , 12 ], the authors utilized CAN-Bus data for driver identification. The car sensors communicated via CAN-BUS (OBD-II protocol), and the CAN-Bus data could be acquired using the OBD-II adapter.…”

“…Among the data obtained from different data sources, CAN-BUS data are the most reliable, feasible, and widely used for driver-behavior profiling [ 10 ]. The security dataset [ 5 ] provides up to 51 features captured using the CAN-BUS data; furthermore, it has been used by several researchers for driver identification [ 7 , 8 ]. In this study, we have used the same security dataset for driver identification.…”

“…However, the continuous connectivity of connected vehicles entails security concerns, rendering cars more vulnerable to hacking and theft [ 5 , 6 ]. This has attracted the continuous attention of researchers toward driver-behavior profiling for identifying [ 5 , 7 , 8 ] the driver for several other applications.…”

“…The in-vehicle controller area network-BUS (CAN-BUS) data include the information corresponding to the steering wheel, vehicle speed, brake-pedal position, and so forth, and the smartphone sensor data include the information related to speed, orientation, and acceleration. Recently, the in-vehicle CAN-BUS data have been regarded as accurate and reliable for driver profiling [ 10 ]; moreover, they have been analyzed for driver identification [ 5 , 6 , 7 , 8 , 12 ] and monitoring [ 9 ]. With the rapid development of the Internet-of-Vehicles technology and the popularization of smart terminal devices such as onboard diagnostic (OBD) devices (the OBD-II protocol provides data from the vehicle ECU ), multi-dimensional CAN-BUS data can be easily captured.…”

“…In this study, we consider the mobile edge computing for a connected car environment and focus on possible embedded solutions that can be deployed in edge server. Moreover, we explore state-of-the-art deep-learning algorithms, namely, fully convolutional networks-LSTM (FCN-LSTM) [ 8 ] and DeepConvRNN-Attention [ 7 ], for driver identification and its feasibility to be implemented in embedded hardware. We evaluated each algorithm with regard to its advantages and disadvantages.…”