Title Page
ABSTRACT
Contens
LIST OF ABBREVIATIONS 12
1. INTRODUCTION 14
1.1. Background 14
1.1.1. Live Video Streaming 14
1.1.2. Peer-to-Peer Video Delivery 16
1.1.3. Mobile Heterogeneous Networks 20
1.1.4. Challenges and Approaches 24
1.2. Motivation 26
1.3. Contributions 31
1.4. Dissertation Organization 33
2. RELATED WORK 34
2.1. RITA 34
2.2. SAMP 36
2.3. Nemati’s 38
2.4. Pan’s 39
3. PROBLEM FORMULATION 42
3.1. Insufficient Path Capacity 43
3.2. Congestive Losses due to Abrupt Bandwidth Changes 48
3.3. Adaptation Delay 50
3.4. Shared Bottleneck Problem 54
4. PATH ADAPTATION SCHEME 60
4.1. Overview 61
4.2. Design Considerations 64
4.2.1. Path Selection at Handoff 65
4.2.2. QoS Preservation during Adaptation 68
4.2.3. Rate Control at Multi-path Handoff 72
4.3. QoS-aware Path Selection with Active Probing 73
4.3.1. Generation of Probing Flow 76
4.3.2. Available Bandwidth Estimation 79
4.3.3. Adjustment of Estimated Bandwidth at SBL 79
4.3.4. Determination of Path Capacities 80
4.4. Conservative Rate Adaptation with Redundancy Control 83
4.4.1. Rate Increase with Path Capacity Information 86
4.4.2. Redundancy Control 87
5. PERFORMANCE EVALUATION 92
5.1. Simulation Model 92
5.2. Simulation Results 98
6. SUMMARY 107
국문요약 110
REFERENCES 112
ACKNOWLEDGEMENTS 124
BIOGRAPHICAL SKETCH 125
Table 1. Comparisons of the related work 41
Table 2. Notation 88
Table 3. System parameter 89
Table 4. Protocol variable 90
Table 5. Protocol messages 91
Table 6. Simulation parameters 95
Table 7. System parameters used in the simulation 97
Figure 1. Example of next generation networks 21
Figure 2. P2P live video streaming over mobile heterogeneous networks 27
Figure 3. Changes in overlay path during handoff 28
Figure 4. Path dynamics 29
Figure 5. Video quality fluctuation 30
Figure 6. Interruption of video playback 30
Figure 7. Tree adaptation algorithm in RITA 35
Figure 8. Multi-homed handoff process in RITA 36
Figure 9. Example of SAMP architecture 37
Figure 10. Sending peer adaptation algorithm of Nemati’s 39
Figure 11. System architecture of Pan’s 40
Figure 12. QoS disruption at path selection 44
Figure 13. Video quality degradation due to insufficient path capacity 45
Figure 14. Playback gap problem due to handoff 46
Figure 15. Example of the playback gap problem 47
Figure 16. Quality degradation due to playback gap 48
Figure 17. Quality degradation due to abrupt bandwidth changes 50
Figure 18. QoS degradation due to adaptation delay 52
Figure 19. QoS degradation due to overlay path selection delay 53
Figure 20. QoS degradation due to rate adaptation delay 54
Figure 21. Shared bottleneck problem 55
Figure 22. Video reception rates during handoff with SBL 58
Figure 23. Example of the network model for a multi-path handoff 62
Figure 24. Flow diagram of the proposed scheme 63
Figure 25. Internal system architecture of the proposed scheme 64
Figure 26. Forwarding capacity 66
Figure 27. Packet pair dispersion 68
Figure 28. Overlapped area for simultaneous binding 69
Figure 29. Horizontal handoff vs. vertical handoff 70
Figure 30. Vertical handoff types 71
Figure 31. Example of the network model for path selection 74
Figure 32. Active probing model 75
Figure 33. Path selection algorithm 76
Figure 34. Generation of probing flow 77
Figure 35. Example of the network model for rate adaptation 84
Figure 36. Rate adaptation algorithm 86
Figure 37. Transit-stub topology generated by GT-ITM 92
Figure 38. Vertical handoff scenario 94
Figure 39. Video reception rates in SP 98
Figure 40. Video reception rates in SP with rate adaptation 99
Figure 41. Video reception rates in SP-PR with rate adaptation 100
Figure 42. Video reception rates in MP-RTT 101
Figure 43. Video reception rates in MP-PR 102
Figure 44. Video reception rates in MP-FC 103
Figure 45. Video reception rates in MP-FC with SBL 104
Figure 46. Video reception rates in MP-PS with SBL 105
Figure 47. Video quality of the proposed scheme with SBL 106