Research And Implementation Of Continuous Identity Authentication System Based On User Behavior

With the popularity of mobile devices,mobile devices contain more and more users' privacy information.The leakage of privacy information will bring huge losses,so the security requirements for mobile devices are higher and higher.The traditional one-time identity authentication method cannot meet the current security needs of mobile devices,so the continuous identity authentication method emerges as the times require.Although the continuous identity method can continuously verify the user's identity,it has the problems of insufficient authentication accuracy and low availability.To solve these problems,this thesis proposes a continuous identity authentication method based on user behavior fusion and a continuous identity authentication method based on context recognition.At the same time,a continuous identity authentication system based on user behavior is designed and implemented.The main research results are as follows:Firstly,in order to solve the problem of insufficient authentication accuracy in user behavior based identity authentication method,this thesis proposes a continuous identity authentication method based on user behavior fusion.First of all,through the experiment,the most favorable touch behavior characteristics to distinguish user identity are determined.Then,we train the authentication model for gait feature and touch feature respectively.Finally,the linear fusion method is used to fuse the output probability score to authenticate the user's identity.The results show that this method can effectively identify whether the user's identity is legal.Our method reduces the error acceptance rate and error rejection rate to 7.22%and 7.14%respectively.Secondly,in order to solve the problem that the usability of user behavior based authentication method is not strong and the authentication accuracy is insufficient,a continuous authentication method based on context recognition is proposed.First,we extract the context recognition features of users to determine the state of user authentication.Then we train the authentication model of different states for each legitimate user.Finally,we match the authentication model of the corresponding state in the authentication process for identity discrimination.The results show that the usability of this method is improved on the basis of effectively identifying the user's identity.Thirdly,an identity authentication system based on user behavior is designed and implemented.The system uses the client server mode,the client is mainly responsible for data collection and upload,the server uses the two identity authentication methods proposed in this thesis to identify the user's identity.