Research And Implement On Multi-domain Policy Integration Based On RBAC

With the development of internet and society, collaboration between multiple organizations has increased drastically. Compared with Web service and grid computing systems which work in multi-domain environment, the collaboration between multiple organizations aims not only for sharing the resources but also for secure access to the resource shared. It is self evident that such collaborations require a formal and secure global policy, which is also easy to administer and consistent with the policies of all the participating domains. Role-Based Access Control (RBAC) model, due to its inherent richness in role hierarchy, separation of duty and least privilege, provides a promising approach to implement a global policy, which satisfies secure requirement in multi-domain environment.In this paper, we propose three attributes of role mapping: transitivity, inheriting type and direction. Role mappings are categorized by the three attributes, which make policy integration more flexible. Then, we proposed a principle for secure role mapping—Permission Decline Principle. The principle describes that role should always be mapped towards permission decline direction. We proposed our RBAC policy integration algorithm based on our role mapping category Permission Decline Principle. Further, we analyze the consistency between global policy and local policy and the resolution of multi-domain policy conflict. Compared with other related algorithms, our policy integration algorithm has better security and performance.We propose a flexible RBAC system software architecture, which separates high level RBAC policy from low level XACML policy. Based on this architecture, a RBAC prototype system has been developed. The system describes RBAC policy under XACML policy language profile, implements RBAC core elements including role hierarchical and separation of duty.We implement our policy integration algorithm in the policy extending layer of software architecture. We utilize the interface in policy abstraction layer implementing main method of global policy, which includes permission set computing, detection of separation of duty, role splitting and role mapping. Further, we demonstrate the correctness and feasibility of our algorithm by an example.