Mar 2005
The Necessity of Supernodes in P2P networks and Distributed Hash Tables

We investigated searching in P2P networks with the goal of showing that supernodes are necessary for efficient searching. We are taking an theoretic approach and aim to establish some results which relate the maximum degree of nodes in a graph with the graphs diameter, which for our purposes, nicely represents the maximum time for a search in a P2P network. We were able to show a diameter of less than the log of the number of nodes in a network is not possible without supernodes, and then construct a graph with diameter of loglog of the number of nodes using a supernodes. This work was a collaboration with David Hadaller and Tyrel Russel.
Feb 2005
Uncertainty & Reputation in Electronic Markets

The domain of reputation modeling in electronic markets has inherent uncertainty due to possible deception on the part of other agents. I am interested in how Bayesian techniques can be applied to capture the probablisitic aspects of the problem and specifically, how belief networks can be used to model how different observations made by an agent can form a single seller reputation and how this can inform the expected quality of a good bought from that seller.
Dec 2004
Desktop Session Mobility - A Device Oriented Solution

This project looks at how we could use a portable device with a large amount of storage, such as an iPod, to store everything necessary to transport a user's entire computing session with them as they move. We implemented a prototype using vmware as a virtual machine which allowed the session to be saved and restored from an iPod. We then examined one approach to how this mobile data could stay sychronized, given a portable device with wireless connectivity. This is joint work with Tyrel Russel.
Nov 2004
Indirect Reputation Assessment of Sellers in Electronic Markets

The Internet allows for A2A commerce at an unprecedented scale; anyone can do business with anyone. The new markets made possible by the Internet bring with them new challenges. This project constructs a system for buyers in electronic markets to avoid bad sellers by modeling the reputation of a seller.