ANALYSIS OF ADAPTIVE STREAMING FOR HYBRID CDN/P2P LIVE VIDEO SYSTEMS

Transcript

1. ANALYSIS OF ADAPTIVE STREAMING FOR HYBRID CDN/P2P LIVE VIDEO SYSTEMS Ahmed Mansy and Mostafa Ammar School of CS, GIT Presented by Tangkai
2. ABOUT THE AUTHOR  Ahmed Mansy  PhD Student  scalable adaptive video streaming over the Internet.  message ferry routing in Disruption Tolerant Networks (DTNs).  Mostafa Ammar  Regents’ Professor & Associate Chair  General Interest: Computer Network Architectures and Protocols.  Current Specific Interests: Overlay Networks, Network Virtualization, Mobile Wirless Networks, Disruption Tolerant Networks.
3. OUTLINE  Introduction  System Description  Single Rate System Model  Adaptive Hybrid Live Video Streaming  Analysis Validation  Illustrative Case Study
4. INTRODUCTION  Video ~ dominate traffic of the Internet.  33% in 2010 ~ 57% in 2014 (expected)  Streaming stored or live video exclude P2P sharing  CDN ~ pillar of the video distribution  Aim: delay and throughput  CDN -> edge server  CDN + adaptive streaming => DASH
5. INTRODUCTION  P2P streaming  280 PT/month in 2009  P2P + adaptive streaming => layered streaming  Cons:  Complicated (design)  High processing power (client)  Not attractive for commercial use  Pros:  Cost-efficiency  CDN/P2P Hybrid System
6. RELATED WORK  Previous works[8][11] on designing such system  LiveSky: operational commercial sys  10m users  1st work study adaptive streaming in CDN/P2P hybrid sys [8] C. Huang, J. Li, , and K. Ross, “Can internet video-on-demand be profitable?” in Sigcomm, 2007. [11] Hao Yin and Xuening Liu and Tongyu Zhan and Vyas Sekar and Feng Qiu and Chuan Lin and Hui Zhang and and Bo Li, “Design and Deployment of a Hybrid CDN-P2P System for Live Video Streaming: Experience with LiveSky,” in Multimedia, 2009.
7. IDENTIFY THE PROBLEM  Assumption  Static in client: no switch/wired ap/constant bw  Dynamic in process: departure and arrival  Bitrate adaption strategy  Linear optimization problem to obtain best suitable bitrate  CDN/P2P mode switch rule  Stochastic fluid model to obtain lower bound of num of user  Interaction between two decision and how they affect each other
8. OUTLINE  Introduction  System Description  Single Rate System Model  Adaptive Hybrid Live Video Streaming  Analysis Validation  Illustrative Case Study
9. DNS REDIRECTION  Typical DNS Lookup  4. Root DNS Server  9. Perfor ming cache
10. SYSTEM DESCRIPTION Core server DNS Redirection Edge server
11. DNS REDIRECTION  [16] [16] A.-J. Su, D. Choffnes, A. Kuzmanovic, and F. Bustamante, “Drafting behind akamai,” in SIGCOMM, 2006.
12. OUTLINE  Introduction  System Description  Single Rate System Model  Adaptive Hybrid Live Video Streaming  Analysis Validation  Illustrative Case Study
13. SINGLE RATE SYSTEM MODEL  Definition  Seeder/leecher  Directly connected to CDN  Unconstrained/constrained  Unlimited number of connections to other peers  Churnless/churn  Fixed number of client  Assumption  Upload rate of all seeder or leecher are the same ( ) l i l u u   ( ) s j s u u  
14. SINGLE RATE SYSTEM MODEL  Unconstrained churnless system  To support r, at least ns seeder   l l s s l l s l l n r u n u n u r n n n     
15. SINGLE RATE SYSTEM MODEL  Unconstrained churn system  Assumption:  User arrival follows Poisson process with rate λ[19]  User stay in sys for a period of time follows general probability distribution with mean 1/γ  Churn happens in leech node only  Total number of user in system N(t) ~ Poisson distribution with rate ρ= λ/γ  Simple admission policy  [19] K. Sripanidkulchai, B. Maggs, and H. Zhang, “An analysis of live streaming workloads on the internet,” in Internet Measurement Conference (IMC), 2004.
16. SINGLE RATE SYSTEM MODEL  Formulation  Poisson distribution(large ρ) -> Gaussian distribution Low bound
17. SINGLE RATE SYSTEM MODEL  Constrained churnless system  Def  Sin number of incoming connection a seeder can accept.(s<-l)  Yin number of incoming connection a leecher can accept.(l<-l)  Yout number of connection leecher can initiate. (l->l+s)  η as the efficiency of the P2P protocol.  Probability leecher can find new content in other leechers.  d as the average download rate for any leecher
18. SINGLE RATE SYSTEM MODEL  = average num of seeder connected to each leecher.   Average leecher download rate is not directly related to the constraints of the system Sin/Yin.  only difference is η with unconstrained churnless sys.
19. SINGLE RATE SYSTEM MODEL  Constrained system with churn  Estimation -> bound  N ~ Gaussian dist( ), (1 − α) confidence interval ,  
20. SINGLE RATE SYSTEM MODEL    is inversely proportional to ρ which means that the higher client arrival rates λ and the longer clients stay in the system 1/γ, the lower becomes.  High guarantee of number of seeder
21. SINGLE RATE SYSTEM MODEL 
22. OUTLINE  Introduction  System Description  Single Rate System Model  Adaptive Hybrid Live Video Streaming  Analysis Validation  Illustrative Case Study
23. ADAPTIVE HYBRID LIVE VIDEO STREAMING  Problem  Which clients should be downgraded to streams of lower bitrates?  What should these new lower bitrates be?  How to get an optimal allocation of bitrates to clients while minimizing client downgrading?  Does the adaptive solution always exist?  Object  client dissatisfaction: difference between bitrate it requested and it actually received  Minimize total client dissatisfaction over all clients.
24. ADAPTIVE HYBRID LIVE VIDEO STREAMING  Unconstrained churnless system  Def:  Bitrates provided by the CDN r1 > r2 > ... > rR  Define xij as the fraction of clients that request bitrate ri but receive bitrate rj 
25. ADAPTIVE HYBRID LIVE VIDEO STREAMING  Linear Optimization problem has a solution. values for xij and nsi  nsi the number of seeders that should receive video of bitrate ri from the proxy.  nsi =0  bitrate ri will not be supported by the server  no clients requested bitrate ri  some clients requested ri but the server decided not to deliver it and downgraded these clients to lower bitrates  nsi >0  does not necessarily mean some clients requested bitrate ri  it could mean that no clients requested rate ri but the server chose to downgrade some of the clients  xij randomly choose fraction of leecher requested ri and delivered rj
26. ADAPTIVE HYBRID LIVE VIDEO STREAMING  Unconstrained churn system  client will request a video stream of bitrate with probability  where λ is the general client arrival rate  number of clients of bitrate at any time in the system becomes a Poisson random variable with an average  Non-linear optimization problem. Use a linear approximation
27. ADAPTIVE HYBRID LIVE VIDEO STREAMING  Constrained churnless system  Constrained churn system
28. ADAPTIVE HYBRID LIVE VIDEO STREAMING  CDN adaptive live streaming  guarantees with confidence (1 − α) that edge server capacity will be sufficient for providing bitrate r to arriving clients with rate ρ.   1 e C r        
29. ADAPTIVE HYBRID LIVE VIDEO STREAMING  CDN v.s. Hybrid Performance  Churnless  Linear optimzation problem -> xij  Churn  approximation
30. OUTLINE  Introduction  System Description  Single Rate System Model  Adaptive Hybrid Live Video Streaming  Analysis Validation  Illustrative Case Study
31. ANALYSIS VALIDATION  Validate single bitrate streaming only  On BitTorrent  Tracker: proxy  Seeder: download torrent and video files  Leecher: download torrent  Parameter  10s chuck  Us/Ul 350kbps/500kbps  ρ 100~400 clients/hour  γ ~ mixed-exponential distribution PDF  Sin = 20, Yin = 10
32. ANALYSIS VALIDATION  Solid line means enough seeder to support bitrate  Fig 4(a) – Fig2(b)
33. ANALYSIS VALIDATION
34. OUTLINE  Introduction  System Description  Single Rate System Model  Adaptive Hybrid Live Video Streaming  Analysis Validation  Illustrative Case Study
35. ILLUSTRATIVE CASE STUDY  Metric  Inter-client fairness  Request and actually received  Saving in CDN server capacity  Profile  low/uniform/high (for bitrate)
36. ILLUSTRATIVE CASE STUDY  Inter-client fairness  Single bitrate manner  Downgrade for all if overloaded.  Adaptive: fairness drop  Single bitrate  Start at lower than 100%/Constant/even better
37. ILLUSTRATIVE CASE STUDY  Capacity saving  Fairness->100%  Saving is less in high profile: asymmetric bw(US/China)
38. Thank you!