Continuous Authentication Based on Keystroke and Mouse Dynamics in Video Private Network

As a typical internal network, the video private network carries various video image resources and involves a large amount of citizen privacy. To prevent potential internal security threats, it is necessary to carry out continuous identity authentication for operators in the video private network. In this paper, a continuous authentication system based on keystroke and mouse dynamics is proposed. The keystroke and mouse operations of users are collected to extract their operating behavior characteristics. An artificial neural network model for each user is constructed for identity authentication. To achieve continuous identity authentication, a novel trust model is present to evaluate users’ identities dynamically. The experimental results show that the proposed method can identify impostors after an average of 115 actions and classify genuine users as impostors only after an average of more than 1,000 actions.