CIF: Small: Fundamental Issues in Peer-to-Peer Communication

University Of Illinois At Urbana-Champaign, Urbana IL

Investigators

Abstract

This project explores fundamental performance and design issues for unstructured peer-to-peer (P2P) communication. It focuses on communication over the public Internet, which is the sharing of widely distributed resources, which are typically owned by different entities, and which typically involve end users' computers acting as both clients and servers. But the analysis also ranges beyond the scope of the Internet, to the ecology of people and devices connected to the Internet. The project contributes to our ability to measure, design, control, and understand the explosion in P2P communication that is already underway. There are exciting possibilities on the horizon for more interactive user generated content, interactive gaming, online interactive instruction, and applications to come. Peer-to-peer communication is likely to significantly alter the structure of the Internet. There is a tension between the open, uncontrolled aura of P2P systems, and reliability and security concerns. By providing a better understanding of the capabilities of P2P mechanisms, this research enhances the value of the future Internet. The investigators study basic aspects of P2P systems in a variety of contexts. These aspects include: strategies of peer selection and piece selection, push vs. pull, proactive vs. reactive, effects of heterogeneous link and host speeds, the impact of incentive mechanisms, effects of network topology--in particular clustering and reduction of inter-ISP traffic, delay, and coding. The contexts include file transfer, live streaming video, video on demand, and delay tolerant interactive systems. The approach is to devise and analyze models capturing various combinations of the above aspects and contexts. The investigators quantify fundamental performance limits, and develop tractable performance analysis and design methods.