Research On Continuous Authentication Integrating Keystroke Content And Keystroke Behavior

Common account password authentication is the first line of defense for network security.With the rapid development of information technology,this authentication method has been unable to meet the current users' security needs.In order to make up for the shortcomings of the account-based password authentication system,verification methods such as dynamic passwords and security cards have appeared,but these physical verifications still have the risk of cracking and leaking.In order to improve the security of user information,biometrics have been used to verify user identity.Biometric identity verification can use fingerprints,irises,voiceprints,etc.These verifications require additional equipment to increase application costs.Therefore,this article analyzes the user's keystroke habits when using the keyboard to confirm the user's identity.This method does not require additional equipment,is not easy to lose,and has high security.The main work and innovation of this article:1)In order to improve the user's continuous identity authentication ability in a real keystroke scenario,this article sets up a completely free experimental environment to collect user keystroke data.This article analyzes the continuous identity authentication under free keystrokes,which does not restrict the user's keystroke content and time,which can meet the authentication requirements of different scenarios and has a wide applicability.2)Considering that the continuous keystroke event can better reflect the user's inherent keystroke habits,the frequency of each subsequent keystroke in the continuous keystroke event is proposed as the keystroke content feature,and the existing keystroke time interval feature Based on this,keystroke behavior characteristics are obtained by sorting them.3)This paper uses the XGBoost algorithm to train the user authentication sub-models for different keystroke domains of the user,and then introduces the improved Yager evidence synthesis theory to fuse the keystroke content domain and keystroke behavior domain sub-models at the decision-making level to obtain a continuous identity authentication model.Experiments show that compared with the existing keystroke authentication model,the authentication method proposed in this paper can effectively improve the accuracy of continuous user authentication,and has higher application value in the real intranet environment.4)This article analyzes the stability of authentication models with different keystroke domains over a longer period of time.Experiments show that the accuracy rate of the authentication model with keystroke behavior domains decreases with time,compared to the keystroke content domains.The authentication sub-model is relatively stable.