Methods for analyzing and preventing Denial of Service (DoS) threats are of fundamental value for designing robust Internet protocols. Much work has been done to develop pragmatic solutions to protocol-specific DoS threats, but there is a lack of realistic theoretical models for studying DoS and of broad paradigms for designing DoS-resilient protocols. This project develops theoretical models based on a "shared channel model" which describes how adversaries and valid senders share the network bandwidth of attack targets. It exploits this model to design counter-measures based on a paradigm in which asymmetries in protocol workloads that are exploited by adversaries are systematically converted to the advantage of trusted parties. Specific project goals include developing (1) general techniques for obtaining DoS-resilience that can be used to adapt existing protocols or create new ones; (2) ways to automate DoS analysis of protocols to reduce the effort required to confirm practical availability properties theoretically and find unexpected attacks before protocols are deployed; and (3) a unified model of integrity, confidentiality, and availability based on both existing algebraic techniques and new probabilistic techniques.
