Research On Authentication Technology Based On User Keystroke Behavior

As people's awareness of privacy increases and the importance of data increases,information security issues become more prominent.As the first protection door for information security,the identity authentication system has become an important breakthrough for cyber attacks.Traditional user authentication focuses on identifying authentication information such as passwords,verification codes,passwords,etc.,and cannot identify the user itself(natural person).Therefore,once the password is stolen or the mobile phone receiving the verification code and password is stolen,its identity is easily impersonated by others.To this end,biometric authentication technology is in demand.Biometric authentication technology is divided into two categories:the use of the inherent physiological characteristics of the human body and the behavioral characteristics of the long-term habits of the human body.Among them,the authentication technologies that utilize the inherent physiological characteristics of the human body mainly include fingerprint authentication,face authentication,and iris authentication.These are more secure than traditional certifications,but these are all one-time verification,that is,they no longer limit their behavior after passing,and lack continuous authentication of behavior.The keystroke authentication technology,which utilizes the long-term habitual behavior characteristics of the human body,can perform continuous identity authentication for users,and has the characteristics of low cost,high availability,easy popularity,and good user interaction.However,the current keystroke authentication has the following problems:First,static authentication can only be used as a login authentication and cannot be used for continuous identity authentication;Secondly,continuous authentication combined with ordinary authentication is often susceptible to tampering attacks due to theft of user passwords.To this end,this paper proposes and designs an identity authentication system based on keystroke behavior,which fully considers the security,convenience,accuracy and continuous authentication functions of the system.The main work of this thesis includes:(1)A static authentication algorithm based on convolutional neural network and a data reorganization method based on this algorithm are proposed.Implementing static identity authentication based on keystroke behavior in the login interface;(2)Propose a continuous authentication scheme based on the weighted reward and punishment mechanism to achieve continuous authentication during the user's use of the computer;(3)Propose an authentication scheme that combines static authentication based on fixed text keystroke features with continuous authentication technology based on free text keystroke features.Static authentication restricts continuous authentication,and continuous authentication improves static authentication,so that the two complement each other;(4)Design and implement an identity authentication system based on keystroke behavior,and carry out functional testing of each module of the system,which can achieve continuous authentication of user identity under the condition of ensuring security and convenience.