Keystroke Authentication with Back Propagation Neural Network

Security threats to computers and networks have increased in considerably because of the increasing use of computer systems and networks in almost every aspect of our daily lives, traditionally, key stroke dynamic authentication is vastly used to authenticate legitimate user in the current system but this method has many loop holes. The aim of this paper is to enhance the continuous authentication method by presenting a keystroke dynamics with back propagation neural network as a transparent layer of user authentication, this biometric technologies is of the most well known and not costly behavior. This paper utilizes keystroke features including Dwell time (DT), Flight time (FT), Up-Up time (UUT), Down-Down time (DDT) and a mixture of these features as strong new keystroke feature. The Back Propagation neural network is used with two cases case-1, used the Back Propagation neural network with sigmoid function, the second case used the Back Propagation neural network with bipolar function, both of these two cases used with five different cases of features as mentioned above. These times are used to distinguish between the authentic users and impostors. Results of the experiments demonstrate that the Back Propagation network with mixed four features in case-1 with mixed of four features is more efficient comparable to case-2 and provide low False Accept Rate (FAR) and False Reject Rate (FRR) and high accuracy