Distinguished Lecture
Arvind Krishnamurthy
Computer Science and Engineering
University of Washington
Thursday, March 12, 2009
3147 MEB
Refreshments 2:00 p.m.
Lecture 2:15 p.m.
Title: Designing Robust Incentive Mechanisms for P2P Systems
Arvind Krishnamurthy
Computer Science and Engineering
University of Washington
Thursday, March 12, 2009
3147 MEB
Refreshments 2:00 p.m.
Lecture 2:15 p.m.
Title: Designing Robust Incentive Mechanisms for P2P Systems
Abstract
A fundamental problem with many peer-to-peer systems is the tendency for users to "free ride"---to consume resources without contributing to the system. The popular file distribution tool BitTorrent was explicitly designed to address this problem, using a tit-for-tat reciprocity strategy to provide positive incentives for nodes to contribute resources to the swarm. While BitTorrent has been extremely successful, we show that its incentive mechanism is not robust to strategic clients. We use these observations to drive the design and implementation of BitTyrant, a BitTorrent client that strategically allocates its bandwidth resources to exploit BitTorrent's incentive mechanism.
Having exploited loopholes in BitTorrent's incentives, and with P2P robustness ultimately dependent on incentivizing users to contribute their resources, we performed a month-long measurement of millions of users to determine the extent to which BitTorrent's incentive mechanism has encouraged user participation. We identify widespread performance and availability problems, surprising given BitTorrent's popularity. These measurements motivate the design and implementation of a new, one hop reputation protocol for P2P networks. Unlike tit-for-tat, where no propagation occurs, or digital currency systems, where contribution information is globally visible, one hop reputations limit propagation to at most one intermediary. Through trace-driven analysis and measurements of a deployment on PlanetLab, we find that one hop reputations can provide wide coverage and positive, long-term contribution incentives.
A fundamental problem with many peer-to-peer systems is the tendency for users to "free ride"---to consume resources without contributing to the system. The popular file distribution tool BitTorrent was explicitly designed to address this problem, using a tit-for-tat reciprocity strategy to provide positive incentives for nodes to contribute resources to the swarm. While BitTorrent has been extremely successful, we show that its incentive mechanism is not robust to strategic clients. We use these observations to drive the design and implementation of BitTyrant, a BitTorrent client that strategically allocates its bandwidth resources to exploit BitTorrent's incentive mechanism.
Having exploited loopholes in BitTorrent's incentives, and with P2P robustness ultimately dependent on incentivizing users to contribute their resources, we performed a month-long measurement of millions of users to determine the extent to which BitTorrent's incentive mechanism has encouraged user participation. We identify widespread performance and availability problems, surprising given BitTorrent's popularity. These measurements motivate the design and implementation of a new, one hop reputation protocol for P2P networks. Unlike tit-for-tat, where no propagation occurs, or digital currency systems, where contribution information is globally visible, one hop reputations limit propagation to at most one intermediary. Through trace-driven analysis and measurements of a deployment on PlanetLab, we find that one hop reputations can provide wide coverage and positive, long-term contribution incentives.
Bio
Arvind Krishnamurthy received his PhD from UC, Berkeley, was on faculty at Yale, and joined UW faculty in 2005. He works primarily at the boundary between the theory and practice of distributed systems and computer networks. His current research interests include peer-to-peer systems, Internet measurements, systems security, and network protocol design.
Arvind Krishnamurthy received his PhD from UC, Berkeley, was on faculty at Yale, and joined UW faculty in 2005. He works primarily at the boundary between the theory and practice of distributed systems and computer networks. His current research interests include peer-to-peer systems, Internet measurements, systems security, and network protocol design.
Return to
2008 Events Calendar