PhD Proposal
Gayatri Swamynathan
Committee: Ben Zhao (co-chair), Kevin Almeroth (co-chair), Rich Wolski
Title: Towards Reliable Reputations for Distributed Applications
Abstract:
The Internet today is experiencing a rapid growth in popularity of online communities like P2P file sharing networks, Wikis, distributed web applications (like Flickr, MySpace, YouTube). The open and anonymous nature of these networks that makes interacting in them so popular also makes them vulnerable to attacks from malicious and self-interested peers. The lack of cooperation among network peers or of a common control managing them aggravate these threats.
Digital reputations are an effective means of social control: they discourage maliciousness and motivate peer cooperation. Designing and deploying a reliable reputation system for large distributed applications, however, remains an open challenge. Current reputation schemes are highly vulnerable to tampering via ratings attacks including false ratings attack, collusion, and Sybil attack. Second, because reputation systems rely on past history to predict future performance, their predictions can be erroneous in high churn systems, where, at any given time, a large portion of peers have little or no history of prior transactions.
The primary focus of this dissertation is to build reputation mechanisms that are resistant to network churn and ratings attacks, and facilitate their integration into distributed applications. In this talk, we address two key challenges posed by short peer lifetimes and peer collusion. We present an overview of proactive reputations, a quick and reliable mechanism to generate reputations for unknown peers and newcomers. To address peer collusion, we propose a reliability metric that estimates the accuracy of a peer reputation based on the number of previous transactions undertaken by the peer and the distribution of these transactions across distinct partners.
Everyone is welcome!