Design And Implementation Of Behavior Authentication System Based On Android System

A study shows that 56% users of smart phone use very simple passwords,or do not use passwords.Therefore,it has practical significance to study a convenient and secure authentication method.With the increasing number of smart phone sensors,more user data can be collected.The authentication system learns the characteristic data generated by the user's use of the behavior,and create a user signature model for constantly user identification,to reach the purpose of refusing to illegal users.Main contents are listed as follows:1.The algorithm of machine learning is studied.By comparing the shortcomings and advantages of all kinds of algorithms,and taking into account the system environment of the algorithm,SVM is chosen as the core algorithm of the system.Finally,through the parameter tuning,data normalization operation,complete the system classifier design.2.A user behavior feature extraction method is proposed,to establish an effective behavior feature set.The behavioral feature set include: the time rhythm,fingertip,and pressure of the click data;the angular sequence,velocity sequence,fingertip sequence,pressure sequence and the maximum,mean,and variance of each sequence of the sliding data;the acceleration data sequence of a gait cycle of and its maximum,mean,variance.3.A method for data preprocessing is proposed.The method completes the preparation of feature extraction by redundant processing,signal de-noising,data classification,coordinate transformation to improve the classification accuracy of the system.4.A comprehensive authentication mechanism is proposed.The mechanism calculate the distance of the user from the historical activity area,and dynamically adjust the system security level;set the Sensitive level of the application,to achieve the purpose of protecting sensitive data.According to the current identification result,it is determined whether to lock the screen and prevent unauthorized usage of user access.5.The principle of data acquisition module,data preprocessing module,feature extraction module,classifier module,authentication module are described in detail.The experiment shows that the behavior authentication system can handle user operation in real time,and has the function of user identification.It can deny the illegal user operation according to the result of user identification.It not only protects the system,but also facilitates the use of the user...