Design and implementation of web-based keystroke analytics for user verification

Stephenson, Ryan (2016) Design and implementation of web-based keystroke analytics for user verification. [USQ Project]

[img]
Preview
Text (Main Project)
Stephenson_R_Zhou.pdf

Download (4Mb) | Preview

Abstract

Keystroke analytics is the study of the way in which a user types rather than simply what they are typing. Through the application of statistical or machine learning methods the gathered biometric data may be used to verify the identity of a user, based on their typing style.

This project aims to explore the field of keystroke analytics to gain an understanding of the methods involved and as such detail the implementation process for such a system’s design and implementation in a web-based context. Details regarding the technical design and implementation are specifically highlighted as current literature often does not describe how the systems shown were developed by rather the theory and methods used by them.

The use of JavaScript to gather typing characteristic data is explored and the process of extracting useful features illustrated. Additionally both PHP and MySQL and used to create the backbone infrastructure to process and store the typing data. A phased development approach has been employed, with the overall system being separated into a collection of subsystems which are designed, implemented and tested before combining them to form the overall system.

The supplementary software system requirements are presented, including the process of setting up a system capable of both being used to perform research on a local system as well as expand to online users for the data collection process.

Method of testing the performance of a keystroke analytics system are discussed with potential changes to improve performance and minimise problems encountered outlined.

The project was successful in that a working proof-of-concept web-based keystroke verification system was designed and implemented which yielded promising results for the data tested (FAR: 0%, FRR: 3.33%). Although to fully evaluate the system’s performance further testing needs to take place for a larger sample size of participants. The results obtained show that a keystroke analytics system may be implemented in a web-based environment, with relatively simple statistical methods, and provide reasonable performance results with only minor additional interaction required by the end-user. This has shown that keystroke analytics is a valid and well-performing method of providing non-intrusive multifactor authentication to traditional login systems.


Statistics for USQ ePrint 31492
Statistics for this ePrint Item
Item Type: USQ Project
Item Status: Live Archive
Additional Information: Bachelor of Engineering (Honours) Major Computer Systems Engineering project
Faculty / Department / School: Current - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering
Supervisors: Zhou, Hong
Date Deposited: 24 Jul 2017 01:34
Last Modified: 24 Jul 2017 01:34
Uncontrolled Keywords: design and implementation; web-based keystroke analytics; user verification; machine learning; biometric data; MySQL; PHP; JavaScript
Fields of Research : 09 Engineering > 0906 Electrical and Electronic Engineering > 090602 Control Systems, Robotics and Automation
URI: http://eprints.usq.edu.au/id/eprint/31492

Actions (login required)

View Item Archive Repository Staff Only