Efficient Data Dissemination in Bandwidth-Asymmetric P2P Networks

As broadband Internet becomes widely available, Peer-to-Peer (P2P) applications over the Internet become increasingly popular. One example of P2P applications, is video multicast in which, a source node streams its video to a large number of destination nodes through an overlay multicast tree consisting of peers. These overlay multicast-based applications, however, do not exploit the full bandwidth of every peer as the leaf nodes in the overlay multicast tree do not contribute their bandwidth to the system. On the other hand, all the peers in a properly constructed overlay mesh can contribute their bandwidth, resulting in high overall system throughput. In this talk, we define the notion of throughput efficiency to measure the performance of different data dissemination schemes from a single source node to multiple destination nodes. Using the proposed throughput efficiency, we propose a hybrid P2P system for optimal data dissemination in a bandwidth-asymmetric network such as Digital Subscriber Line (DSL) access network. Our hyprid P2P system consists of one or more Supernodes responsible for node management and a large number of streaming nodes, Peers which are responsible for the actual data delivery. Our system is designed to achieve (1) bandwidth fairness, i.e. a node receives the same amount of bandwidth that it contributes to the system, (2) small end-to-end delay, (3) small node management’s overhead, and (4) optimal bandwidth usages of all the nodes. Both simulations and experimental results of a prototype system consisting of PlanetLab nodes demonstrate the aforementioned qualities.

Speaker Details

Nguyen earned a B.S. from the University of Washington, and an M.S. and Ph.D. from U.C. Berkeley in 2000 and 2003, respectively. His doctoral work focused on multimedia streaming over the Internet, using the path diversity framework in conjunction with network protocols, and source and channel coding techniques. This work won him best paper award at Packet Video Workshop 2002. His current research interests include computer networks, signal processing, machine learning, and data mining.Before joining Oregon State University in September, 2004, he was a post-doctoral research associate in the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory. While at Lawrence Livermore National Lab, Nguyen involved in a number of projects ranging from analyzing statistics of streamed dataset to image processing. From 1998 to 2003, he was a research assistant in the Electrical Engineering and Computer Science at UC Berkeley. From 1996-1998, he was a graphics researcher at Intel’s Microcomputer Research Lab, working on fast methods for 3-D visualization and navigation in human body. He also spent some time at Microsoft, optimizing DirectX6 for Pentium chips. From 1995 to 1996, he worked on caches for multi-processor system in the Microprocessor Group at Intel, Corp. Before that, he worked on PBX switches at ROLM, Siemens.

Date:
Speakers:
Thinh Nguyen
Affiliation:
Oregon State University
    • Portrait of Jeff Running

      Jeff Running