Improving resiliency using capacity-aware multicast tree in p2p-based streaming environments

  • Authors:
  • Eunseok Kim;Jiyong Jang;Sungyoung Park;Alan Sussman;Jae Soo Yoo

  • Affiliations:
  • Department of Computer Science and Interdisciplinary Program of Integrated Biotechnology, Sogang University, Seoul, Korea;Department of Computer Science and Interdisciplinary Program of Integrated Biotechnology, Sogang University, Seoul, Korea;Department of Computer Science and Interdisciplinary Program of Integrated Biotechnology, Sogang University, Seoul, Korea;UMIACS and Department of Computer Science, University of Maryland;School of Electrical and Computer Engineering, Chungbuk National University, Cheongju Chungbuk, Korea

  • Venue:
  • HPCC'06 Proceedings of the Second international conference on High Performance Computing and Communications
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a capacity-aware multicast tree construction scheme for P2P-based live streaming environments (R-CAT). The R-CAT builds resilient multicast trees by considering the capacity of participating nodes and locating the high capacity nodes toward the upper parts of the trees. The resulting trees become shallow and more resilient since the number of affected nodes is minimized when nodes leave or fail. We have implemented R-CAT on top of Splitstream, a high bandwidth streaming system using Pastry, and evaluated our scheme in terms of packet loss rate and end-to-end delay. The performance results show that R-CAT is more resilient and provides lower end-to-end delay under various group dynamics.