Authentication of Smartphone Users Using Behavioral Biometrics

Abstract

This research investigates the application of behavioral biometrics for secure smartphone authentication. The primary objective is to develop a robust and user-friendly system that surpasses traditional password-based methods. The scope includes data acquisition, feature extraction from typing rhythm and touch patterns, and the development of a machine learning-based authentication model. The conclusion demonstrates that behavioral biometrics offer a viable alternative, achieving high accuracy while maintaining user convenience, thus addressing the limitations of existing authentication methods.

Introduction

Smartphone security is paramount, given the sensitive personal data stored on these devices. Traditional password-based authentication is vulnerable to phishing and shoulder surfing. Biometric methods, like fingerprint or facial recognition, while secure, can be bypassed or spoofed. Behavioral biometrics, leveraging inherent user behaviors like typing rhythm and touchscreen interactions, offer a compelling alternative. However, challenges remain in achieving high accuracy while minimizing user inconvenience and ensuring robustness against adversarial attacks. This research aims to address these challenges by developing a novel authentication system combining multiple behavioral features.

Objectives

• To develop a robust behavioral biometric authentication system for smartphones.
• To achieve high accuracy and low error rates in user authentication.
• To design a user-friendly system that minimizes inconvenience for legitimate users.

Technical Details

• Behavioral data acquisition from keystroke and touch dynamics
• Feature extraction: timing intervals, pressure, swipe direction
• Machine learning classifiers (e.g., SVM, Random Forest, KNN)
• Performance metrics: accuracy, precision, recall, FAR/FRR
• Development Tools: Android Studio, Python, Scikit-learn