Continuous user identity verification using typing error classifications

Abstract

Computer systems are to be used in our daily life and security threats are seen increasing as ever. Authentication checks the identity of a person whenever he/she logins to the system. Continuous authentication is used to identify or verify if the legitimate user is present until the session time out. Continuous authentication using keystroke and mouse dynamics are already proposed, which requires additional hardware to record the biometric features. But this work proposes a continuous authentication using typing or spelling errors in keystroke dynamics. Many types of errors will come in the case of keyboard typing. Four types of errors are used here: Dynamic, diagonal, interpolation and contralateral errors. In addition to these four errors the character corrected with backspace are also considered. These errors have least frequency error rate compared with other errors. So these are considered as keystroke features. This method can be easily deployed and provides accuracy in continuous authentication.