P2P and Online Social Networking Research at Mirage Group

In this talk, I present an overview of three projects in my research group.

1. The PRIME project focuses on fundamental issues in the design and evaluations of P2P streaming mechanisms. I present a few studies in the context of this project including (i) design of a scalable mesh-based P2P streaming mechanism for scalable delivery of live video, (ii) a comparative study between mesh-based and tree-based approaches to live P2P streaming, and (iii) the importance of packet scheduling schemes in mesh-based P2Pstreaming mechanisms.
2. The goal of IonP2P project is to investigate and develop new measurement and modeling methodologies to understand and accurately characterize properties and dynamics of large scale P2P systems. First I show that the key challenge in accurately characterizing large scale P2P systems is to capture their representative (i.e. unbiased) “snapshots”. I demonstrate how different factors can introduce bias into the selection of representative samples of peer properties (e.g. degree, link bandwidth). Then, I present design and evaluation of a couple of theoretically-grounded techniques for unbiased sampling of dynamic graphs. I provide a summary of the empirical characterizations of several unstructured and structured P2P systems.
3. The third project that we recently started, tackles the empirical characterization of online social networks (OSN). I discuss the motivations and challenges in characterizing OSNs, then sketch an array of issues that we plan to explore in this area and present some of our preliminary results.