Affiliations: University of Texas at Dallas Richardson, TX 75083, USA | Purdue University, West Lafayette, IN 47907, USA
Abstract: Even if Role Based Access Control (RBAC) is employed properly, distributed database environments are still prone to illegitimate access attempts: in RBAC, users potentially carry the risk of illegal access attempts via credentials violation, or unintentional/intentional incorrect use of already granted permissions via role misuse/abuse. We introduce a probabilistic risk management model for enhanced access control in such databases. We incorporate failure modes and effects analysis (FMEA) scheme for measuring user risks in our design. We combine components as credentials, queries, role history logs and expected utility for a probabilistic formulation of risk. We present experimental results that we obtained on real world database. The results emphasize the need for a database where roles are well defined and queries under different roles do not overlap. We suggest using query templates with minimized role definitions so that the risk model we introduce provides better risk management.