User authentication based on keystroke dynamics using artificial neural networks

Other Title(s)

التحقق من المستخدم استنادا على ديناميكية ضغطة المفتاح باستخدام شبكات عصبية اصطناعية

Dissertant

Hubi, Mays Muhammad

Thesis advisor

Hamid, Sarab Majid

Comitee Members

Duaymi, Mahdi Kazaz.
Stephan, Jane Jalil
Abd al-Wahhab, Halah Bahjat

University

University of Baghdad

Faculty

College of Science

Department

Department of Computer Science

University Country

Iraq

Degree

Master

Degree Date

2012

English Abstract

Computer systems and networks are being used in almost every aspect of our daily life, the security threats to computers and networks have increased significantly.

Usually, password-based user authentication is used to authenticate the legitimate user.

However, this method has many gaps such as password sharing, brute force attack, dictionary attack and guessing.

Keystroke dynamics is one of the famous and inexpensive behavioral biometric technologies, which authenticate a user based on the analysis of his/her typing rhythm.

In this way, intrusion becomes more difficult because the password as well as the typing speed must match with the correct keystroke patterns.

This thesis considers static keystroke dynamics as a transparent layer of the user for user authentication.

Back Propagation Neural Network (BPNN) and the Probabilistic Neural Network (PNN) are used as a classifier to discriminate between the authentic and impostor users.

Furthermore, four keystroke dynamics features namely: Dwell Time (DT), Flight Time (FT), Up-Up Time (UUT), and a mixture of (DT) and (FT) are extracted to verify whether the users could be properly authenticated.

Two datasets (keystroke-1) and (keystroke-2) are used to show the applicability of the proposed Keystroke dynamics user authentication system.

The best results obtained with lowest false rates and highest accuracy when using UUT compared with DT and FT features and comparable to combination of DT and FT, because of UUT as one direct feature that implicitly contained the two other features DT, and FT; that lead to build a new feature from the previous two features making the last feature having more capability to discriminate the authentic users from the impostors.

In addition, authentication with UUT alone instead of the combination of DT and FT reduce the complexity and computational time of the neural network when compared with combination of DT and FT features.

Main Subjects

Information Technology and Computer Science

Topics

• Contracts
• Scientific innovations
• Data processing
• Computers
• Access control
• Neural networks(Computer science)

No. of Pages

118

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-605390