Graduate Research Forum Details
Cheating in Peer-to-Peer Cycle Sharing Systems
| Authors: | Shanyu Zhao Yuhong Liu |
|---|---|
| Date: | April 26, 2004 |
| Time: | 16:00 |
| Location: | 220 Deschutes |
Abstract
Peer-to-peer computing, the harnessing of idle compute cycles throughout the Internet, offers exciting research challenges in the converging domains of networking and distributed computing. The NRG's research project, Cluster Computing on the Fly (CCOF) seeks to harvest idle cycles from ordinary users in an open, untrusted environment. The CCOF model of cycle sharing is more general than that used in SETI@home and other well-known cycle sharing systems; it supports four classes of computations: infinite workpile, workpile with deadlines, tree-based search, and point-of-presence applications.
In this seminar, we will present the problem of verifying correctness of results in an untrusted cycle-sharing environment. We are investigating three proposed solutions: replication, stand-alone quizzes, and embedded quizzes.
After giving background about CCOF and the results verification problem, we seek your input on two questions:
- What kinds of cheating can occur in a cycle sharing system?
- How hard is it to embed secret quizzes and how hard is it for a malicious node to detect embedded quizzes, in application binaries or source code?
We especially invite the programming language and software engineering students to come brainstorm with us on these issues.