Role-Based Access Control (RBAC) defines structured authorization policies by associating permissions with roles rather than individual users. While RBAC enforces formal access constraints, it does not ensure that role usage remains behaviorally consistent over time. In low-resource IT environments, continuous monitoring of role integrity is often limited to log retention without analytical oversight. This work proposes a lightweight, role-aware deviation modeling framework designed to identify policy integrity drift using standard authentication and application logs. Instead of framing deviation as malicious activity, the approach conceptualizes divergence as measurable inconsistency between expected and observed role behavior. The framework models baseline activity across frequency, temporal, and structural dimensions and computes interpretable deviation scores without reliance on deep learning architectures or labeled attack datasets. Evaluation using controlled scenarios demonstrates that the proposed method detects meaningful role-based drift while maintaining computational efficiency and stable alert volume. The framework operates with linear complexity and minimal infrastructure requirements, making it suitable for deployment in resource-constrained environments. This study contributes a governance-oriented perspective to access control research by bridging formal policy design and operational behavior visibility