Biometric patterns recognition using keystroke dynamics

Resumo

This paper aims to describe a strategy for biometric authentication embedded system that uses keystroke dynamics to recognize the users. The main motivation of this work is a gap identified on the biometric authentication devices market that demonstrates the lack of a low cost and high efficiency product. Therefore, the use of low cost microcontrollers coupled with a good biometric authentication strategy could fill this gap. The PIC and ESP microcontrollers were used to create a prototype with the purpose of performing measurements and generating users' biometric models. During these measurements 9 volunteers had their typing characteristics extracted and stored. After data collection, several tests were performed and values of 36% for FRR and 7.2% for FAR were found. More expensive results can still be achieved by modifying some punctualities in data collection, as commented at the end of the paper.