CAREER: Advances in Peer-to-Peer Networking

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

Metcalfe's law states that the "value" of a network increases with the square of the number of users. However, for most applications, as the number of participants increases, performance degrades, and so too does the value of application to the user. Centralized service and application architectures are most at risk for not maintaining performance with an increasing user base. Centralized sites can represent a processing bottleneck, a single point of failure, or single point ofattack. Furthermore, a single server cannot mitigate for client's poor network conditions in third-party domains. Repli- cating and distributing servers is one solution for solving the problems of centralized architectures. Peer-to-peer services take distribution even further by using collaborative clients who are pre- pared and expected to perform some or all the functions of servers. As new clients bring new load to the systems, they also bring new resources. Thus, peer-to-peer, collaborative services can be a powerful paradigm: if on average clients bring enough or more than enough resources to balance the load and cost they incur, then peer-to-peer services can increase in value and performance with the number of participants. The primary purpose of this proposal is to make fundamental advances in the understanding and design of applications and services that make use of replication and collaboration in a peer-to-peer environment. Opportunities for the application of peer-to-peer techniques in new disciplines seem limitless. I propose to examine several applications that are only possible, or achieve their best performance, using peer-to-peer architectures and not centralized architectures. These include the novel peer-to-peer applications of anonymity, secure ad-hoc routing, mobility-based transport for wearable computers, cheat-proof serverless gaming, and distributed information retrieval. There are commonalities to all the numerous peer-to-peer applications we will study: the ex- istence of heterogeneous peer resources, redundancy of offered services at peers, and replication of application data and content. By addressing these issues, the researcher will make important advances in discovery and understanding of peer-to-peer networking and by applying those results to a diverse set of problem areas, the researcher's work will make a broader impact in in fields as diverse as security (privacy, availability, and secure routing), mobile computing, wearable computing, multimedia and gaming, and distributed information retrieval. The major theme of the researcher's proposed educational activities is the involvement of undergraduates in original research. The goal of the educational proposal is to prepare undergraduate students for graduate school by providing them with an early opportunity toperform research. It consists of an integrated plan of publishable undergraduate research projects, new course curriculum, the development of practical lab exercises in networking and network security, distance learning, and minority outreach. By focusing the research of undergraduates on the themes of this proposal, the researcher is ensuring a number of benefits. First, that the impact of this educational component is tightly integrated with my research objectives. Second, by coordinating efforts with the director of the Northeast Alliance for Graduate Education and the Professoriate (NEAGEP), the researcher will improve minority outreach in the department by preparing and encouraging under-represented minority students to continue on for doctoral programs at UMass and other universities. Third, the researcher will advance the department's teaching mission by providing a new three-course curriculum of networking and security. Fourth, the researcher will produce a standard set of practical laboratory exercises in networking and network security to be used at any university. Finally, the availability ofmy curriculum and practical exercises, as well as UMass's existing Video Instruction Program and MANIC distance learning programs, will increase the external impact of the curriculum.