Resumen
Role-based access control (RBAC) is one of the most popular access-control mechanisms because of its convenience for management and various security policies, such as cardinality constraints, mutually exclusive constraints, and user-capability constraints. Role-engineering technology is an effective method to construct RBAC systems. However, mining scales are very large, and there are redundancies in the mining results. Furthermore, conventional role-engineering methods not only do not consider more than one cardinality constraint, but also cannot ensure authorization security. To address these issues, this paper proposes a novel method called role-engineering optimization with cardinality constraints and user-oriented mutually exclusive constraints (REO_CCUMEC). First, we convert the basic role mining into a clustering problem, based on the similarities between users and use-partitioning and compression technologies, in order to eliminate redundancies, while maintaining its usability for mining roles. Second, we present three role-optimization problems and the corresponding algorithms for satisfying single or double cardinality constraints. Third, in order to evaluate the performance of authorizations in a role-engineering system, the maximal role assignments are implemented, while satisfying multiple security constraints. The theoretical analyses and experiments demonstrate the accuracy, effectiveness, and efficiency of the proposed method.