Behavioral Trust Drift Detection (BTDD): A Lightweight Framework for Dynamic Authentication

AbstractAuthentication systems traditionally evaluate users based on current risk scores or predefinedauthentication rules. However, these approaches often fail to capture sudden behavioral changesthat may indicate malicious activity. This paper proposes Behavioral Trust Drift Detection(BTDD), a lightweight authentication framework that evaluates the variation in user trust overconsecutive authentication sessions. The proposed model introduces a Behavioral Trust Drift(BTD) metric that measures the difference between current and previous trust states. A dynamicauthentication decision is then performed based on the calculated drift value. The framework isvalidated using the CIC-IDS2017 dataset, where network flows are interpreted as authenticationattempts. Experimental results demonstrate that BTDD achieves improved detection performancecompared with traditional static and risk-based authentication methods. The proposed frameworkoffers a simple, scalable, and behavior-aware approach for dynamic authentication environments.