Research On Continuous Identity Authentication Based On User Behavior

Identity authentication is a process in which the computer and network system authenticate the operator's digital identity.It plays an important role in the protection of computer network security.The computer can only recognize the digital identity of users in cyberspace,and authorize the digital identity of the user.Once the user's digital identity is stolen,all resources and privileges in his network account will be exploited by malicious users.Continuous identity authentication is the continuous and periodic detection of users' identity when they visit.There have some security problems that the traditional identity authentication scheme based on static key and dynamic password only authenticates users once when they log in.Aiming at the problem of identity counterfeiting in the process of user network interaction,this thesis proposes a continuous identity authentication scheme based on user behavior characteristics.Using web browsing behavior with distinct individual characteristics and behavior patterns to describe users' identities can improve the security of information systems and prevent counterfeit users from stealing digital identities.The main research contents and contributions are as follows:1.According to the distribution of the number of pages visited by users on time points,the concept of fuzzy time window is introduced to represent the real time,and the appropriate size of time window is selected for periodic authentication of users' access behavior.2.In the process of feature extraction,in order to reduce the spatial dimension of sequence feature vectors to mine frequent sequence patterns with time series,a maximum proportion pruning strategy based on stack is proposed to mine the maximum frequent sequence set.3.The feature subspace is constructed by fusing multiple features,and the user continuous identity authentication model is built based on multi-layer perception network in the optimized feature subspace.It can identify the user's identity after the user visits several webpages.If the result is abnormal,it interrupts the user's operation and carries out static identity authentication,and updates the model according to the result of authentication.Experiments on data sets show that this method has highrecognition rate in the process of continuous detection,and the recognition rate of recognition model changes with time is more accurate.In this thesis,25 users are tested for identity authentication.The results show that the accuracy of the proposed scheme reaches 98.3% and the recall rate reaches 98.4%.We can authenticate user's identity after them visiting ten pages.