The current Internet has reached all areas of computing and communication from government to the enterprise and from the enterprise to the home. However, the Internet falls short of current expectations for a robust, secure communication infrastructure and future demands of network-centric operations. Perhaps the attributes most critically lacking are those related to the availability of network services. Our research will explore new network architectures and operational procedures that can provide service access guarantees while under attack at all levels of communication. We are motivated by the fundamental observation that present-day adversaries (e.g., operating inexpensive or even free “bot” networks) are beginning to target the network infrastructure itself (e.g., links, specific routers, and router clusters), not just end services and hosts. Yet none of the proposed countermeasures to these attacks have provided for dependable access to legitimate traffic during such attacks. In fact, we have recently demonstrated, via Internet-scale simulated attacks, that infrastructure-targeting flooding attacks could effectively cut off over 90% of the Internet traffic to an enterprise of the size and connectivity of CMU, over 50% of the traffic to a US state like Virginia, and over 30% of the traffic to the West Coast of the US.
In addressing infrastructure threats, an overarching goal is to create new network architectures and operational procedures that provide high levels of availability, even in environments with a large number of malicious end-hosts and network operators. Our expectation is that research on how to construct networks offering exceedingly high levels of availability will enable us to make network-centric operations substantially more available than they are today. We will also plan to initiate new research in the application of trustworthy computing technologies to network infrastructures, aiming to increase the security and availability of future networks.