Abstract:
Privacy disclosure is a big problem when smart mobile devices lost. Now, the biometric technologies must rely on the corresponding auxiliary equipment, such as fingerprint recognition, face recognition, resulting in the authentication is complex and costly. To solve these problems, a gait biometrics-based mobile device authentication solution is proposed in this paper. In the training phase, the gait data of different behaviors of users in daily life are collected by the accelerometer in the mobile device. Then, the feature is extracted to construct feature vector and establish the gait model of users. The model matching algorithm based on the neural network is used to achieve the purpose of identification during the identification phase. The C/S architecture is applied to the implementation of the system. In order to ensure the security of network data transmission, all the transmission data is encrypted by SMS4 symmetric encryption algorithm. Lots of experiments show that the average recognition rate of neural network algorithm is 78.13 percent, and integrating the feedback mechanism, the authentication accuracy can up to 98.96 percent.