Multi-phase mouse dynamics authentication system using behavioural biometrics
Pooja S Pilankar, Puja Padiya · 2016
In this era, the password-based system is not enough to secure important data. It has, in general, the disadvantage that passwords are either too hard to remember or too easy to guess. It is necessary to apply a higher level of security. This can be achieved by using the biometrics i.e. behavioral biometrics such as using mouse dynamics. In the existing system, the user needs to perform a specified task with his mouse pointer, and based on the performance of that task, the user is granted or denied access to the computer. In this paper, we are proposing a method, where we provide additional mouse dynamics to the existing work i.e. mouse single and double click related tasks and combination of which is later used as their identity. Our intention is to improve the accuracy of the system by adding more features of the mouse for easily classifying the genuine and impostor user. The mouse data obtained from the user is raw data and it has to be normalized and make outlier free to used for the further process so this is resolved by using Peirce's criterion and Weighted Least Square Regression (WLSR) and for classification, we have used Learning Vector Quantization(LVQ).