Keystroke dynamics based authentication using fuzzy logic

With online information thefts increasing every day, enhanced user authentication systems have become a necessity. This is due to the fact that current authentication systems have certain weaknesses, such as the username used for login is easily known to anyone and password can be guessed or forcibly retrieved. This paper focuses on overcoming these weaknesses by recognizing users based on their typing patterns. This technique which is used to uniquely label users based on their typing rhythm is known as keystroke dynamics. In an authentication system employing keystroke dynamics, samples provided by the users during the initial sign up phase are used to generate a template for the future authentication. This template is regularly updated whenever the user is correctly identified. The work proposed in this paper focuses on implementing a keystroke dynamics based authentication system using fuzzy logic. It also discusses how a fuzzy logic approach is beneficial as compared to other approaches implemented previously, and how a continuous learning model improves accuracy over time. The proposed fuzzy logic model gives 98% authentication accuracy and 2% false rejection rate.