Analysis of Strong Password Using Keystroke Dynamics Authentication in Touch Screen Devices

in this paper user verification and identification system on touch screen mobile devices is proposed. The system examines the keystroke dynamics and uses it as a second authentication factor. The study proposes a prototype for a keyboard application developed for collecting timing and non-timing information from keystroke dynamics. In addition to other mentioned in literature studies, we propose complex password combination, which consists of text, numbers, and special characters. Strengthening access control using artificial neural networking model is suggested. Neural network model based on multilayer perceptron classifier which uses back propagation algorithm is proposed. This paper presents a unique approach for combining timing and non-timing features together, as it includes several non-timing features such pressure, size, and position in addition to the duration time features. Several experiments have been done based on specific machine learning for data mining and classification toolkit named WEKA. The obtained results show that keystroke dynamics provides acceptable level of performance measures as a second authentication factor. The distinguishable role for non-timing features beside the timing features is demonstrated. These features have a significant role for improving the performance measures of keystroke dynamic behavioral authentication. The proposed model achieves lower error rate of false acceptance of 2.2%, false rejection of 8.67%, and equal error rate of 5.43% which are better than most of references provided in the literature.