Keystroke Biometric Studies with Short Numeric Input on Smartphones

Michael J Coakley, Pace University

Abstract

A keystroke biometric system was developed to accommodate touchscreen keystroke features. In addition to the usual key-press and key-transition features available from mechanical keyboards, the touchscreen features also included pressure, press location, accelerometer, and gyroscope information. Short numeric input data, ten-digit numeric strings, were collected from 52 participants on identical Android smartphones. A number of user authentication biometric experiments were performed on these data to measure overall system performance and to quantify the biometric value of various feature subsets. System performance was measured using the standard biometric techniques of receiver operating characteristic curves and equal error rates. Two validation models (repeated random sub-sampling and leave-one-out cross-validation) and two distance metrics (Euclidean and Manhattan) were compared in the study. Results were compared to previous keystroke biometric studies. The touchscreen keystroke features not available on mechanical keyboards greatly outperformed the usual mechanical-keyboard-like keystroke features with an equal error rate of 3.9% to 19.7%, respectively. Additionally, the Manhattan distance metric outperformed the Euclidean distance metric, and leave-one-out cross-validation outperformed repeated random sub-sampling. Of the various touchscreen feature subsets, the gyroscope features performed best with an equal error rate of 4.3%.^

Subject Area

Computer science

Recommended Citation

Michael J Coakley, "Keystroke Biometric Studies with Short Numeric Input on Smartphones" (January 1, 2016). ETD Collection for Pace University. Paper AAI10117796.
http://digitalcommons.pace.edu/dissertations/AAI10117796

Share

COinS

Remote User: Click Here to Login (must have Pace University remote login ID and password. Once logged in, click on the View More link above)