1. Opportunities and Challenges of Peer-to-Peer Internet Video Broadcast Speaker: Shao-Fen Chou Adivisor: Dr. Ho-Ting Wu 11/14/2012 1

2. Outline Introduction Peer-to-peer video broadcast Case studies Technical challenges and open issues Deployment status and challenges Summary Reference 2

3. Introduction In recent years, there has been significant interest in the use of peer-to-peer technologies for Internet video broadcast. Two key drivers make the approach attractive: (1) Such technology does not require support from Internet infrastructure. (2) A participant is not only downloading a video stream but also uploading it to others. 3

4. Introduction Peer-to-Peer technologies have emerged as important for a wide range of applications such as file download and voice-over-IP. The distinguishing and stringent requirements of video broadcast necessitate fundamentally different design decisions and approaches. 4

5. Peer-to-peer video broadcast Contrast from other peer-to-peer applications A video broadcast system typically has a single dedicated source. There are several distinguishing characteristics of such a system: (1) Large scale (2) Performance-demanding (3) Real-time constraints (4) Gracefully degradable quality 5

6. Peer-to-peer video broadcast Design issues There are important criteria for overlay construction and maintenace. (1) Overlay efficiency (2) Scalability and load balancing (3) Self-organizing (4) Honor per-node bandwidth constraints (5) System consideration 6

7. Peer-to-peer video broadcast Approaches for overlay construction We focus on the approach taken towards the overlay structure used for data dissemination. In particular, the proposals can be broadly classified into two categories: (1) Tree-based approach (2) Data-driven randomized approach 7

8. Peer-to-peer video broadcast Approaches for overlay construction Tree-based approach: Peers are organized into structures for delivering data. This approaches are typically push-based. One concern with tree-based approaches is that the failure of nodes. Loop avoidance is an important issue that must be addressed. 8

9. Peer-to-peer video broadcast Approaches for overlay construction Data-driven randomized approach: Data-driven overlay designs do not construct and maintain an explicit structure. A approach to distributing data is to use gossip algorithms. Some approaches adopt pull-based techniques. 9

10. Case studies Example Tree-Based Approach: ESM The ESM system employs a structure-based overlay protocol that is distributed, self- organizing, performance-aware, and constructs a tree rooted at the source. The tree is optimized primarily for bandwidth and secondary for delay. 10

11. Case studies Example Tree-Based Approach: ESM Group management Each ESM node maintains information about a small random subset of members. A new node joins the broadcast by contacting the source and retrieve a random list of members that are currently in the group. To learn about members, a gossip-like protocol is used. 11

12. Case studies Example Tree-Based Approach: ESM Membership dynamic When node leaves, members continue forwarding data for a short period. Performance-aware adaptation Each node maintains the throughput it is receiving in a recent time window. Detection time indicates how long a node must stay with a poorly performing parent. 12

13. Case studies Example Tree-Based Approach: ESM Parent Selection Each node B that responds provides information about: (1) the performance (2) whether it is degree-saturated (3) whether it is a descedant of A A switches to the parent B either if (1) the estimated throughput of B is high enough for A (2) B improves delay 13

14. Case studies Example Resilient Structure Approach: Multitrees Two key advantages of the multitree solution: (1) The overall resiliency of the system is improved. (2) The potential bandwidth of all nodes can be utilized. 14

15. Case studies Example Resilient Structure Approach: Multitrees An example of multitree 15

16. Case studies Example Data-Driven Approach: CoolStreaming A CoolStreaming node consists of three modules: (1) a membership manager (2) a partnership manager (3) a scheuler 16

17. Case studies Example Data-Driven Approach: CoolStreaming Group and parent management It requires newly joining nodes to contact the origin server. It employs an existing scalable gossip membership protocol to distribute membership messages. A video stream is divided into segments, and the avalibilty of the active segments in the buffer of a node is represented by a buffer map(BM). 17

18. Case studies Example Data-Driven Approach: CoolStreaming Group and parent management An illustration of partnership in CoolStreaming, with A being the source node 18

19. Case studies Example Data-Driven Approach: CoolStreaming Scheduling Algorithm CoolStreaming uses a sliding window to represent the active buffer portion. A BM consists of a bit string of 120 bits, each indicating the availabilit of the corresponding segment. 19

20. Case studies Example Data-Driven Approach: CoolStreaming Scheduling Algorithm The scheduling algorithm strikes to meet two constraints: (1) the playback deadline for each segment (2) heterogeneous streaming bandwidth from the partners 20

21. Case studies Example Data-Driven Approach: CoolStreaming Failure recovery and partnership refinement The departure can be detected after an idle time. CoolStreaming lets each node periodcally establish new partnership. This operation serves two purpose: (1) maintain a stable number of partners (2) explore partners of better quality 21

22. Technical challenges and open issues Tree-based versus data-driven: could there be any hybrid? The key challenge is that a set of stable nodes needs to be positioned at appropriate locations. It may conflict with the bandwidth and delay optimization in tree construction. 22

23. Technical challenges and open issues Tree-based versus data-driven: could there be any hybrid? 23

24. Technical challenges and open issues Incentives and fairness There could be many free riders in a peer-to-peer system. CoopNet assume each node contributes as much badwidth as it receives. BitTorrent-like applications adopt a tit-for-tat stategy to solve the incentive problem. 24

25. Technical challenges and open issues Incentives and fairness 25

26. Technical challenges and open issues Access bandwidth scarce regimes A key challenge is that the nodes behind DSL and cable can receive several hundreds of kilobits per second but can fundamental only donate less. Using additional nodes not in the peer-to-peer system, called waypoints. 26

27. Technical challenges and open issues Extreme peer dynamics and flash crowd The system has to rapidly assimilate the new peers without significant impacting the video quality of existing and newly peers. Designing a peer-to-peer video broadcast system that is robust to extreme peer dynamics is still an open research problem. 27

28. Technical challenges and open issues Support for heterogeneous receivers Video is encoded at multiple bitrates in parallel and is broadcast simultaneously. Recent proposals leverage another specialized coding algorithm called multiple description coding(MDC). 28

29. Technical challenges and open issues Network coding: coding at peers The fundamental insight in network coding is that if data can be encoded in intermediate nodes, then the optimal multicast troughput can be achieved. 29

30. Technical challenges and open issues Network coding: coding at peers 30

31. Technical challenges and open issues Implementation issues NATs and Firewall Transport protocol Startup delay and buffer interaction 31

32. Deployment status and challenges Deployment status With higher user participation, the statictical results are even better. Deployment challenge The key challenge pertains to the conclicting interests faced by network and content service providers and the differences between how the Internet and the traditional video content providers operate. 32

33. Summary Among the three video distribution modes: broadcast, on-demand streaming, and file download, broadcast is the most challenging to support. Peer-to-peer solutions represent the most promising technical approaches for Internet video broadcast due to te self-scaling property. 33

34. Reference J. Liu, S. G. Rao, B. Li, and H. Zhang, "Opportunities and challenges of peer-to-peer internet video broadcast," Proceedings of the IEEE, vol. 96, no. 1, pp. 11-24, 2008. 34