Date of Award

2019

Document Type

Open Access Dissertation

Degree Name

Doctor of Philosophy in Computer Science (PhD)

Administrative Home Department

Department of Computer Science

Advisor 1

Jean Mayo

Advisor 2

Chaoli Wang

Committee Member 1

Ching-Kuang Shene

Committee Member 2

Steven Carr

Abstract

Attacks on computers today present in many different forms, causing malfunction of operating systems, information leakage and loss of business and public trust. Access control is a technique that stands as the last line of protection restricting the access of users or processes to resources on computers. Throughout the years, many access control models have been implemented to accommodate security requirements under different circumstances. However, the learning of access control models and the management of access control policies are still challenging given its abstract nature, the lack of an environment for practice, and the intricacy of fulfilling complex security goals. These problems seriously reduce the usability of access control models.

In this dissertation, we present a set of pedagogical systems that facilitates the teaching and studying of access control models and a visualization system that aids the authoring and analysis of access control policies. These systems are designed to tackle the usability problems in two steps. First, the pedagogical systems were designed for new learners to overcome the obstacles of learning access control and the lack of practicing environment at the very beginning. Contrary to the traditional lecture and in-paper homework method, the tool allows users to write/import a policy file, follow the visual steps to understand the concepts and access mechanisms of a model and conduct self-evaluation through Quiz and Query modules. Each of the four systems is specifically designed for a model of the Domain Type Enforcement, Multi-level Security, Role-based Access Control, or UNIX permissions. Through these systems, users are able to take an active role in exploring the effect of a policy with a safe and intact underlying operating systems. Second, writing and evaluating the effect of a policy could also be challenging and tedious even for security professionals when there are thousands of lines of rules. We believe that writing an access control policy should not include the complexity of learning a new language, and managing the policies should never be manual when automatic examination could take the place. In the aspect of policy writing, the visualization system kept the least number of key elements for specifying a rule: user, object, and action. They describe the active entity who takes the action, the file or directory which the action is applied to, and the type of accesses allowed, respectively. Because of its simple form without requiring the learning of a programming-like language, we hope that specifying policies using our language could be accomplished effortlessly not only by security professionals but also by anyone who is interested in access control. Moreover, policies can often be left unexamined when deployed. This is similar to releasing program which was untested and could lead to dangerous results. Therefore, the visualization system provides ways to explore and analyze access control policies to help confirm the effect of the policies. Through interactive textual and graphical illustrations, users could specify the accesses to check, and be notified when problems exist.

Share

COinS