iCar: Multi-modal Learning-Inspired Vehicle Owner Recognition Based on FMCW Radar in Internet of Things

The existing human behavior recognition algorithms based on wireless signals are easily affected by the surrounding environment or weather. The built-in FMCW (Frequency Modulated Continuous Wave) radar in the vehicle can collect dynamic biometric features of people, which can be used to verify the identity of the car owner while accomplishing its predetermined tasks. In this paper, we propose a novel multi-modal learning-inspired vehicle owner identification and authentication frame-work, called as iCar, which integrates time-frequency features and micro-Doppler features. In particular, these two features are abstracted from original time-frequency spectrograms and micro-Doppler spectrograms, which are applied to handle frequency changes over time and dynamic biometric features, respectively. Experimental results and performance analysis confirm the superiority of the proposed algorithm.