Three-layer authentication in keystroke dynamics using time based tool

Abstract

Having a secure and sound system is the most important need of the end-user. Confidential and authentic information about the system must be available to the genuine user when required. This paper presents a generalized Keystroke Dynamics Technique for identification of genuine users. The method works for the authentication of the user while user is entering the password to using the system. It is a three-layer approach which first check the typing pattern while entering the password then it also monitors the system while user is using the system. Different users have different typing pattern which could be used to recognize a user. For identification of the user a time-based tool is used to collect data pertaining to the typing time of each user for words of different lengths. This is a very easy and cost-effective way of collecting data for differentiating between a genuine user and imposter. At first layer elbow method is used to know unknown targets depending upon different word combinations. For the second layer principal component analysis (PCA) is used to find suitable factors where user typing pattern is making users indistinguishable. For the third step Long-Short Term Memory (LSTM) technique is used to forecast whether the user is a genuine user or an imposter.