Getting access control policies right is challenging, especially in large organizations. This project is developing techniques and tools to support efficient and trustworthy administration of Attribute-Based Access Control (ABAC) policies. ABAC is a flexible, high-level, and increasingly popular security policy framework.
ABAC promises long-term cost savings through reduced administrative effort, but manual development of an initial ABAC policy can be expensive. This project is developing policy mining algorithms that promise to drastically reduce the cost of migrating from legacy access control frameworks to ABAC. These algorithms generate candidate ABAC policies from existing lower-level policies, if available, or operation logs, together with data about attributes of users and resources.
An administrative policy specifies how each user may change the access control policy. Fully understanding the implications of administrative policies in enterprise systems can be difficult, because of the size and complexity of the policies, and because sequences of changes by different users may interact in unexpected ways. This project is developing policy analysis algorithms that support validation of administrative policies, by answering questions such as whether, how, and under what conditions specified administrators can together change the policy in order to grant specified permissions to specified users.
Powerful development environments for creating and validating access control policies, incorporating algorithms like the ones being developed in this project, have the potential to significantly increase the trustworthiness of IT systems, by helping security administrators efficiently and reliably develop correct policies.