Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

BitTorrent Protocol


Published on

Location Awareness in
Unstructured Peer-to-Peer

Published in: Education
  • Be the first to comment

  • Be the first to like this

BitTorrent Protocol

  1. 1. Location Awareness in Unstructured Peer-to-Peer Systems IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 16, NO. 2, FEBRUARY 2005 KuYou Lin 2005/05/26
  2. 2. Outline Introduction Related Work Location-Aware Topology Matching Three Main Operations Simulation & Performance Evaluation Conclusion
  3. 3. 1.Introduction LTM (Location-aware Topology matching) Unstructured P2P Solve Mismatching problem Distributed , Doesn’t require global view Main idea Build an efficient overlay by disconnecting slow connection Choosing closer peer as neighbor
  4. 4. 2.Related Work Other ways to reduce traffic cost in unstructure-P2P system Forwarding-based Only forward to subset of neighbors Cache-based Remember index of files/peers used before Overlay topology optimization Logical topology / physical topology LTM
  5. 5. 3.Location-Aware Topology Matching Problems at unstructured P2P system Unnecessary Message Duplications in Overlay Connections Topology Mismatch (logical/physical)
  6. 6. Unnecessary Message Duplications in Overlay Connections
  7. 7. Topology Mismatch C1 C2 C1 C2 C3 C4 C3 C4 C1 C4 C2 C3
  8. 8. 4.Three Main Operations TTL2-Detector Flooding Flooding 2 hops Slow Connection Cutting Source Peer Probing
  9. 9. TTL2-Detector Flooding Each peer floods a TTL2-Detector periodically d (i, S, v) i : Message ID S: Initial Source v: TTL value d (i, S, 1) Source Peer One hop Peer d (i, S, 0)
  10. 10. Slow Connection Cutting 3 kinds of cases Will-Cut List / Cut List - d (i, S, 1) SP d (i, S, 0) SN - - NP 1 1
  11. 11. Source Peer Probing Peer P receives only one d(i,s,0) during a certain time P - (N2(S)-N(S)) C
  12. 12. Definition: N(S) N2(S) S
  13. 13. Operation Example
  14. 14. 5.Simulation & Performance Evaluation Performance Metrics Average traffic cost vs. search scope Tc = Message * number of Links Search scope = number of peers reached Average neighbor distance Query response time
  15. 15. Simulation Environment Overlay (logical): 2000,3000,5000,8000 nodes Physical: 22000 Internet-like nodes Neighbors: 4,6,8,10 neighbors
  16. 16. LTM in Static Environment Traffic cost vs. Search scope (Only 1 step LTM)
  17. 17. Traffic cost vs. optimization step
  18. 18. LTM in Dynamic Environment Effectiveness of Will-Cut List (W-c-50 50 sec) 95% 70%
  19. 19. Effective ness of cut list LTM-k means k-times LTM / per minute
  20. 20. Traffic cost of 4 schemes 57 13
  21. 21. Average response time of 4 schemes 24 6
  22. 22. 6.Conclusion Using LTM in unstructured P2P system can reduce 75% traffic cost and 65% query response time Will-cut List and cut List can improve the performance of LTM LTM is completely distributed and scalable
  23. 23. Reference IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 16, NO. 2, FEBRUARY 2005 Yunhao Liu, Member, IEEE, Li Xiao, Member, IEEE, Xiaomei Liu, Lionel M. Ni, Fellow, IEEE, and Xiaodong Zhang, Senior Member, IEEE