Enhancing smartphone security with human centric bimodal fallback authentication leveraging sensors

Area of Science:

• Computer Science
• Cybersecurity
• Human-Computer Interaction

Background:

• Smartphones contain sensitive personal data, requiring advanced security measures.
• Existing authentication methods may not offer sufficient protection against sophisticated threats.

Purpose of the Study:

• To develop and evaluate a lightweight bi-model fallback authentication system for smartphones.
• To enhance user data protection by combining dynamic security questions with inertial measurement unit (IMU) data.

Main Methods:

• Dynamic security questions were generated based on user smartphone behavior (calls, SMS, app usage, etc.).
• Finger movement patterns were captured using accelerometer, gyroscope, gravity sensor, and magnetometer (IMUs).
• A bi-model system combined question responses and IMU data for authentication, tested with 24 participants over 28 days.

Main Results:

• Dynamic security questions, particularly those based on call, SMS, and app usage, showed high accuracy.
• Integrating IMUs significantly boosted authentication accuracy from a maximum of $$ to $$.
• The True Positive Rate improved from 0.79 to 0.99, demonstrating enhanced detection capabilities.

Conclusions:

• The proposed lightweight bi-model fallback authentication is effective for smartphone security.
• Combining dynamic security questions with IMU data offers a robust and improved user authentication solution.
• This technique presents a promising advancement in securing personal information on mobile devices.