Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 Selection Strategies for Peer-to-Peer 3D Streaming Wei-Lun Sung, Shun-Yun Hu, Jehn-Ruey Jiang National Central University, Taiwan 2008/05/29.

Similar presentations


Presentation on theme: "1 Selection Strategies for Peer-to-Peer 3D Streaming Wei-Lun Sung, Shun-Yun Hu, Jehn-Ruey Jiang National Central University, Taiwan 2008/05/29."— Presentation transcript:

1 1 Selection Strategies for Peer-to-Peer 3D Streaming Wei-Lun Sung, Shun-Yun Hu, Jehn-Ruey Jiang National Central University, Taiwan 2008/05/29

2 National Central University, Taiwan 2 Virtual environments (VE) VEs allow users to interact in synthetic worlds Larger content & more worlds  content streaming (i.e., 3D streaming) becomes necessary

3 National Central University, Taiwan 3 3D streaming Continuous and real-time delivery of 3D content to allow user interactions without a full download.  Object streaming fragments mesh into base & refinements Base123 Refinements User (Hoppe 96)

4 National Central University, Taiwan 4 Scene streaming multiple objects object selection & prioritization [Teler & Lischinski 2001]

5 National Central University, Taiwan 5 Comparison with media streaming Highly interactive (latency-sensitive) Behavior-based (non-linear) How to scale to millions of concurrent users?

6 National Central University, Taiwan 6 Imagine you start with a globe

7 National Central University, Taiwan 7 Zoom in…

8 National Central University, Taiwan 8 To a city

9 National Central University, Taiwan 9 and a building

10 National Central University, Taiwan 10 Right now it’s flat…

11 National Central University, Taiwan 11 But in the near future…

12 National Central University, Taiwan 12 Observation Limited & predictable area of interest (AOI)‏ Overlapped visibility = shared content

13 National Central University, Taiwan 13 Benefits of peer-to-peer Scalable  Growing amount of total resources Affordable  Commodity PCs Feasible  Better client hardware (CPU, broadband networks)‏  Availability of user-hosted machines

14 National Central University, Taiwan 14 Peer selection Choose suitable candidates so that content retrieval can be done quickly and efficiently Source discovery  Which peers possess the needed data Source selection  Which peers to request the data

15 National Central University, Taiwan 15 Related Work: FLoD [Infocom 2008] VE partitioned into cells with scene descriptions Assumes P2P overlay that provides AOI neighbors star: selftriangles: neighbors circle: AOIrectangles: objects

16 National Central University, Taiwan 16 Peer selection in FLoD Source discovery  Query-response  Extra delay due to queries Source selection  Random selection  Requests contention due to overlapping requests

17 National Central University, Taiwan 17 OBJ Request contention problem Overlapping requests create contentions R1 R2 R3 R4 R5 R6 R1,R2 R1,R2,R3 R1,R2,R3, R4,R5,R6

18 National Central University, Taiwan 18 Proposed Solutions

19 National Central University, Taiwan 19 Incremental Piece List Exchange Proactive notification of content availability Periodic incremental exchange of content availability information with neighbors. Msg_TypeObj_IDMax_PIDObj_IDMax_PID ‧‧‧‧ incremental content information

20 National Central University, Taiwan 20 Extended Candidate Buffer Non-AOI neighbors may still possess data Maintain extra list of non-AOI neighbors R S Obj

21 National Central University, Taiwan 21 Multi-Level AOI Request Localized requests may prevent contentions Peers request from closer neighbors/levels first

22 National Central University, Taiwan 22 Simulation Environment Based on FLoD (available on SourceForge) World size: 1000 x 1000 Simulation steps:3000 Objects: 500 Nodes: 50 ~ 500 (50 nodes increase) AOI radius: 75 Server bandwidth: 10 Mbps / 10 Mbps Peer bandwidth: 1 Mbps / 256 Kbps

23 National Central University, Taiwan 23 Simulation Environments (cont.) Source discovery  (QR) query-response: 5 steps interval, 10 requests  (EE) exchanged & extended: 150 radius Source selection  (RAND) random  (ML)multi-level AOI request : 4 levels Original FLoD: QR-RAND Proposed method:EE-ML

24 National Central University, Taiwan 24 Hit Ratio

25 National Central University, Taiwan 25 Base Latency

26 National Central University, Taiwan 26 Fill ratio

27 National Central University, Taiwan 27 Bandwidth (Server)

28 National Central University, Taiwan 28 Bandwidth (Clients source discovery)

29 National Central University, Taiwan 29 Conclusion New selection strategies for P2P 3D streaming  Availability info exchange & extended candidate buffer reduce both latency and bandwidth overhead  multi-level AOI requests obtain data from closer providers but improve only hit ratio Future work  More sources  Physical topology  Pre-fetching

30 National Central University, Taiwan 30 Q & A

31 National Central University, Taiwan 31 Neighbor discovery via VON Boundary neighbors New neighbors Non-overlapped neighbors [Hu et al. 06] Voronoi diagrams identify boundary neighbors for neighbor discovery

32 National Central University, Taiwan 32 LODDT ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ ‧ Object Tree NodeAura

33 National Central University, Taiwan 33 LODDT ‧ ‧ ‧ ‧ ‧ Object Tree NodeAura U

34 National Central University, Taiwan 34 LODDT (cont.) Discovery  Estimation Selection  Every peer samples the time-to-serve (TTS) of its neighbors  Requestors organize their data requests so as obtain tree nodes in the right order Drawback: incorrect estimation, congestion Requests Candidates

35 National Central University, Taiwan 35 Simulation Environments (cont.) System performance  Hit ratio: Ratio of successful requests peers have sent  Latency: Duration between initial request and data arrival  Fill ratio: Ratio of the possessed required data Scalability metrics  Bandwidth usage (consumption)  Content discovery overhead


Download ppt "1 Selection Strategies for Peer-to-Peer 3D Streaming Wei-Lun Sung, Shun-Yun Hu, Jehn-Ruey Jiang National Central University, Taiwan 2008/05/29."

Similar presentations


Ads by Google