Different Internet users often access similar content resulting in identical data repeatedly traversing the network. Several systems have attempted to eliminate this redundant data from network links to improve network performance. However, these only apply to point-deployments, i.e. to specific applications, protocols, or links. Because of this, their benefits are limited.
This project develops an Internet architecture where redundancy elimination (RE) is inherently supported by the network. This architecture subsumes localized deployments and extends the benefits of RE to all end-to-end flows. Crucially, it improves the design of applications and protocols. It frees application designers from concerns about using network resources efficiently. Network protocols can leverage inherent RE to better meet application and network constraints. This work will consider the technical, algorithmic and economic issues in this new architecture, including supporting RE in networks and end-hosts, reconfiguring applications to optimally leverage RE, and incentives for ISPs and end-hosts to deploy RE.
Intellectual merit: The new architecture will improve end-to-end performance, enhance applications and protocols, and give rise to new applications. The research will consider technical, algorithmic, and economic issues in supporting RE and leveraging it optimally. It will advance various fields of Computer Science, including network hardware, protocols and applications, algorithms, game-theory and network economics.
Broader impact: This project will have a broad impact on future applications and protocols. It will improve network management and routing. It will also impact ISP pricing. The project has a substantial educational component involving the introduction of new courses and cross-site mentoring.
