The Application of Keystroke Analysis for Physical Security: A Field Experiment

Abstract

A major factor restricting the widespread acceptance of biometric security technologies is the lack of experimental results from repeated use in a real world setting. Specifically, with respect to keystroke analysis, previous research has primarily discussed the classification performance of the biometric by controlling different variables during laboratory experiments. This research is an extension of previous work that investigates the classification performance of keystroke analysis using a biometric keypad and several novel typing features during a 5-week field experiment. The ultimate goal is to explore stabilization patterns over time and to test if field study results support the optimal parameters and classification rates identified in the laboratory study. Results from the experiment demonstrate that a classification rate of 87% can be achieved during actual use of the biometric keypad by employing the optimal parameters. This finding helps to validate and extend previous work conducted in a laboratory setting. It also provides useful insight into experimental designs and support for the application of keystroke analysis in a real world environment.