The Research And Implementation Of User Authentication Scheme Based On Mobilephone Sensor

With the popularity and rapid development of smart phones, the functions and businesses involved are becoming more and more widely. Most mobile phones use a password or pattern to protect users’ private information at present. But these authentication mechanisms are easy to be invaded and unable to prove the user’s real identity. Therefore, a more security authentication methods for mobilephone is needed. Based on the 3-D gesture signature, a new authentication system is proposed to solve the authentication issues of mobilephone. And a prototype system has been finished, which is designed and developed on smart phone with Android platform. The main contributions of this dissertation are:(1) An method research of 3-D gesture signature authentication based on SVM classifier. The original gesture acceleration data are collected by accelerometer embedded in the mobilephone. To cope with the noise interference of raw data, research has been done on data preprocessing techniques, including smoothing data by simple moving average method, normalize data by the mean-standard deviation normalization, and the data of the moving gesture are intercepted by using a threshold value to retain effective data. Speed and displacement values are obtained by integrating the acceleration data. All the data’s time domain characteristics are extracted, such as mean value and variance. Through SVM classifier to complete the authentication. The simulation results have shown that the algorithm is efficiency.(2) An method research of 3-D gesture signature authentication based on dynamic matching. To solve the problem of mismatch on the timeline between different acceleration data of 3-D gesture signatures and the difficulty of quantifying the differences, an algorithm of aligned sequences interpolation is proposed. Absolute distance is used to quantify the differences between different gestures, and through threshold method judging user’s authenticity. The simulation results have shown that the algorithm has better recognition performance contrast with the other two existing similar algorithms.(3) An method research of 3-D gesture signature authentication based on D-S evidence theory and template renewing. After the research of 3-D gesture signature authentication based on dynamic matching, the D-S evidence theory is used for authentication to improve the stability of results. On the one hand, a basic distribution function model is established, and a distribute authentication method is used for fusion the recognition results of accelerometer and gyroscope. Meanwhile, use the fused result as a final result of authentication. On the other hand, a template renewing strategy based on cluster analysis is proposed to keep the method long-term effectiveness. Simulation results have shown the algorithm is efficiency.(4) A system research of 3-D gesture signature authentication based on Android platform. The development of the 3-D Gesture Signature Authentication System(3-DGSAS) on Android platform has been done. Install this application on different brands of mobile phones, an experimental test has been finished with 19 volunteers, and the total test times is 76,520. The experimental results have shown that the average result of users false acceptance rate(FAR) is 0.31% and average result of users false rejection rate(FRR) is 4.33%.