A theory for comparing the expressive power of access control models

Comparing the expressive power of access control models is recognized as a fundamental problem in computer security. While such comparisons are generally based on simulations between different access control schemes, the definitions for simulations that are used in the literature are informal, and make it impossible to put results and claims about the expressive power of access control models into a single context. Furthermore, some definitions for simulations used in the literature such as those used for comparing RBAC (Role-Based Access Control) with other models, are too weak to distinguish access control models from one another in a meaningful way. We propose a theory for comparing the expressive power of access control models. We perceive access control systems as state-transition systems and require simulations to preserve security properties. We discuss the rationale behind such a theory, apply the theory to reexamine some existing work on the expressive power of access control models in the literature, and present four results. We show that: (1) the well known HRU scheme is limited in its expressive power when compared to a rather simple trust-management scheme, thereby formally establishing a conjecture from the literature; (2) RBAC with a particular administrative scheme from the literature (ARBAC97) is limited in its expressive power, countering claims in the literature that RBAC is more expressive than DAC (Discretionary Access Control) schemes; (3) the ability to check for the absence of rights (in addition to the presence of rights) causes ATAM (Augmented Typed Access Matrix) to be more expressive than TAM (Typed Access Matrix); and (4) a trust-management scheme is at least as expressive as RBAC with a particular administrative scheme (the URA97 component of ARBAC97).