Network routing is the ensemble of technologies for determining optimal paths for data delivery. These paths are selected as a result of complex policy-driven interactions between multiple networks and protocols. It is hard to understand the consequences of adopting particular policy configurations, especially since each individual operator's view of the global network is incomplete. Some combinations of policy can lead to loss of connectivity, the adoption of sub-optimal paths, or protocol convergence failure.
This research project aims to provide theoretical and practical methods by which operators can manage the interaction of policy in the absence of complete information. The central formalism is a protocol-agnostic model of partial policy specifications. This model supports a variety of tasks, such as the synthesis of unspecified parts of local policy according to high-level objectives, configuration repair, and planned migration from one policy to another. Each of these applications is to be given a rigorous algorithmic basis in terms of the policy model, and implemented on simulated and real routing platforms. Additional research goals include the development of higher-level policy representations on top of the basic model, and providing tools for assessing policy privacy using the incomplete information framework.
The broader impacts of this research include the development of practical tools for policy analysis and synthesis, on standard network platforms. This will allow important routing policy management tasks to be carried out with confidence. Student projects connected to the development and validation of these tools will be of direct importance to their education.
