SemanticAction: A Semantic-Driven and Behavior-Aware Password Framework for Adaptive VR Authentication Under...

Area of Science:

• Human-computer interaction
• Cybersecurity
• Virtual Reality

Background:

• Traditional authentication methods like passwords face memorability issues and are vulnerable to observation attacks (e.g., Man-in-the-Room).
• Biometric and behavioral authentication in VR raise practical, privacy, and deployment concerns.
• Secure and usable authentication is crucial for the increasing adoption of immersive Virtual Reality (VR) applications.

Purpose of the Study:

• To introduce SemanticAction, a novel authentication framework for VR.
• To integrate knowledge-based passwords with gesture-based behavioral biometrics.
• To address security vulnerabilities and usability challenges in VR authentication.

Main Methods:

• SemanticAction encodes passwords as scene-anchored directional gestures, leveraging semantic meaning for intuitive interaction.
• The framework employs randomized scene prompts and decoy scenes to mitigate observation attacks.
• A dual-constraint verification mechanism is utilized to adapt to varying data availability (cold-start/few-shot).

Main Results:

• Two user studies indicate SemanticAction effectively mitigates Man-in-the-Room (MITR) attacks.
• The framework alleviates password memorability challenges.
• Favorable usability and security performance were observed, even with limited behavioral data.

Conclusions:

• SemanticAction offers a promising approach to behavior-aware authentication in VR.
• The framework provides practical design considerations for future VR security systems.
• Integrating semantic-driven actions with behavioral biometrics enhances VR authentication security and usability.