Authenticating User's Keystroke Based on Statistical Models
Ying Zhang, Guiran Chang, Lin Liu, Jie Jia · 2010
In this paper, we use statistical methods to establish a keystroke biometrics model to authenticate a user's identity by predicting the user's keystroke behavior characteristics. We use HMM for keystroke sequence analysis and time series to compute the state output probability of HMM used in keystroke biometrics model. At the authentication phase, we use modified forward algorithm to compute the users' typing behavior state. We also collect the users' keystroke data to establish the authentication model. Then using fixed text analysis and digraph's keystroke duration time, we implement the authentication mechanism. Extensive experiments have verified the effectiveness of the proposed solutions.