Haptic-Based Sensible Graphical Password

Authentication, the process of identifying an individual, currently plays a critical role in security systems. Graphical password schemes are believed to be more secure and resilient to dictionary attacks than traditional textual passwords. We designed a graphical password system with the novelty of incorporating the sense of touch via haptic technologies. Using haptics, the system utilizes the physical attributes captured during human-haptic-computer interactions. Parameters such as pressure and velocity can be handled as ‘hidden’ features to increase the resiliency of the system. Initial testing showed that the users do not follow the conventional password’s guidelines, thus making the system easy to be flawed. Consequently, we proposed two solutions: enforce stringent conditions during password creation, or modify the verification algorithm. The use of haptics for password creation lends itself ideally to the latter solution and to a subsequent modification of the authentication algorithm. Therefore, we adapted the initial pressure recording system to authenticate individuals by using a two-class classification Artificial Neural Network (ANN) and Nearest-Neighbor’ (NN) algorithms. On average, NN approach allowed for a (92.0 ± 0.3) % probability of successful verification. On the other hand, by classifying the ‘hidden-feature’ vectors using an ANN approach, the probability of successful password verification becomes (90.09 ± 3.8) %. Overall, both methods offer similar Probability of Verification (PV) rates.