An Enhanced User Privacy Model in Context Aware Authentication System using Behavioural Biometrics

Authentication is a fundamental component of digital security, yet traditional mechanisms such as passwords and tokens are increasingly vulnerable to compromise. While behavioural biometrics and context-aware authentication have emerged as promising alternatives, they introduce privacy concerns when sensitive data is stored or transmitted without adequate protection. This paper presents a privacy-preserving, context-aware authentication model that integrates behavioural biometrics with contextual intelligence and Elliptic Curve Cryptography (ECC) to enhance both security and user privacy. The proposed system captures keystroke dynamics as the behavioural trait, augments them with contextual parameters such as device location, network type, and time of access, and applies ECC to ensure end-to-end confidentiality of sensitive features. Experimental evaluation using a benchmark keystroke dataset augmented with simulated contextual information demonstrated high authentication accuracy (96.8%), low Equal Error Rate (2.7%), and excellent ROC-AUC (0.982). Statistical tests confirmed that the proposed model significantly outperforms a biometrics-only baseline (p < 0.01), while operational metrics showed negligible computational overhead (~20 ms increase in authentication time). Error analysis indicated that contextual checks reduced successful impostor attempts by 35% compared to behavioural biometrics alone. These results highlight the feasibility of combining behavioural biometrics, contextual awareness, and ECC to deliver a secure, privacy-compliant, and real-time authentication solution suitable for deployment in high-security domains such as finance, healthcare, and cloud-based services.