Alliance-Based Peering Scheme for P2P Live Media Streaming

1. An Alliance based Peering Scheme for P2P Live Media Streaming Darshan Purandare Ratan Guha University of Central Florida August 31, 2007 @ P2P-TV, Kyoto

2. Outline • Chunk based P2P models • Current Issues • Proposed peering scheme • Discussion and summary Darshan Purandare

3. Chunk based P2P models … 1 … Server 3 2 … … … 5 4 … … 1 … • Stream is split into pieces 3 Darshan Purandare

4. Working Philosophy • Peers exchange buffer map of data availability • Retrieve the missing pieces and upload pieces to neighbors • Scheduling algorithm finds which pieces to retrieve and from whom • Partnership refinement helps to obtain better peers in terms of uploading rate Darshan Purandare

5. Outline • Chunk based P2P models • Current Issues • Proposed peering scheme • Discussion and summary Darshan Purandare

6. Current Issues • Quality of Service can improve [Hei et al. 06] • Long start up time • Peer Lag • Unfairness [Ali et al. 06] • Uplink bandwidth distribution uneven • Sub-optimal uplink utilization • May affect QoS & Scalability • Can we do better ? Darshan Purandare

7. Outline • Chunk based P2P models • Current Issues • Proposed peering scheme • Discussion and summary Darshan Purandare

8. Proposed Model • Chunk based paradigm but overlay formation using alliances • Nodes cluster in groups of 4-8 to form alliances • Media server relays content to Powernodes • BEAM: Bit strEAMing Darshan Purandare

9. BEAM: Working Philosophy • A new node upon arrival obtains peerlist from Tracker • Peerlist contains nodes in similar bandwidth range and similar network (if possible) • Contacts peers for stream content • Starts joining alliances or creates one • Server relays stream content to Power nodes • Power nodes changes periodically based on Utility Factor (UF) • A node’s UF computed using: • Cumulative share ratio (CSR) • Temporal share ratio (TSR) Darshan Purandare

10. Alliance Formation Peerlist of Node 6: 12, 22, 43 Peerlist of Node 1:: 6, 17, 23 Darshan Purandare

11. Alliance Properties • A node can be a member of multiple alliances • H: Maximum number of nodes in an Alliance • K: Maximum number of alliances a node can join • Number of neighbors = K(H-1) Darshan Purandare

12. H = 5 K = 2 Alliance Functionality Darshan Purandare

13. H = 5 K = 2 Alliance Functionality Darshan Purandare

14. Why form Alliances ? • Clustering into alliances forms a small world network graph • Short Path Length • Robust to network perturbations such as churn • Close knit group ensures streaming content is readily available within alliances • Alliance members have common trust & treaty Darshan Purandare

15. Small World Network • Dense local clustering (high clustering coefficient) • Some links to other part of the graph (non local) • Overlay distance is near-optimal • Robust to churn • [Watts et al., Nature,98] Darshan Purandare

16. Simulator Details • Streaming rate = 512 Kbps • Media Server’s Uplink = 1536 Kbps (3 links) • Heterogeneous bandwidth class • (512,128), (768,256), (1024, 512), (1536,768), (2048, 1024) • H, K = 4, 2 (6 neighbor nodes) • Each node buffers content for 120 sec Darshan Purandare

17. QoS: Average Jitter Rate Darshan Purandare

18. QoS: Average Latency Darshan Purandare

19. Uplink Utilization Darshan Purandare

20. Fairness: Share Ratio Range Darshan Purandare

21. Fairness: Jitter Factor Range Darshan Purandare

22. Fairness: Latency Range Darshan Purandare

23. Summary • Alliance based peering scheme is an effective technique to group peers • QoS, Uplink throughput and fairness results are at par or even better than CoolStreaming • Peer lag can be improved using BEAM • Initial buffering time can be slightly improved • Our research is complementary to advance source and channel coding techniques. Darshan Purandare

24. Questions or Comments Darshan Purandare

25. Thank You!