Privacy-Preserving Continuous Authentication for Zero-Trust Remote Work Environments Using Multimodal Behavioral Biometrics and Cancellable Hashing
Contributors
BANGARU SINDHU
G SIVA SANKAR REDDY
Keywords
Proceeding
Track
General Track
License
Copyright (c) 2026 Sustainable Global Societies Initiative

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Abstract
The attack surface of corporate systems has expanded due to remote employment and login-only authentication cannot detect session hijacking or changes in user activity during a session. Continuous behavioral authentication addresses this gap. However, it suffers from large intra-user variability and the privacy risk of storing biometric templates. In this paper, we propose Privacy-Preserving Multimodal Continuous Authentication (PMCA) for Zero-Trust remote offices based on keystroke dynamics and mouse behavior. Behavior-Preserving Augmentation captures natural within-user variation. Modality-Specific Transformer Encoders learn temporal patterns Cross-Modal Behavioral Attention captures typing-mouse interaction Adaptive Weighted Graph Fusion builds a unified representation. A Cancellable Hashing Transformation transforms this representation into a revocable, user-keyed protected template. A continuously updated trust score, calculated on protected representations, informs a Zero-Trust policy that allows access, requests step-up verification or ends the session. We evaluate on two public datasets, BEHACOM and the Behavior Biometrics Dataset, using accuracy, Equal Error Rate, irreversibility, unlinkability, revocability and compromise-resistance measures. No empirical results have been reported yet. The PMCA seeks to provide safe, privacy-conscious continuous verification in enterprise remote employment.