1. 1 On a Unified Architecture for Video-on-Demand Services Jack Y. B. Lee IEEE TRANSACTIONS ON MULTIMEDIA, VOL. 4, NO. 1, MARCH 2002

2. 2 Outline Introduction UVoD Architecture Performance Modeling Numerical Results Simulation Results Interactive Controls Conclusions

3. 3 Introduction true-VoD (TVoD) Service quality is maximized near-VoD (NVoD) System cost is minimized unified VoD (UVoD) Cost-performance tradeoff

4. 4 UVoD Architecture (1)

5. 5 UVoD Architecture (2)

6. 6 UVoD Architecture (3)

7. 7 Admit-via-Unicast Arrives at time t t m-1 < t < (t m – δ ) After (t – t m-1 )

8. 8 Recourse reduction over TVoD Admit-via-Multicast As multicast channels is fixed, Admit-via-Multicast users will not result in additional load Increasing the admission threshold δ then more user will be admitted to the multicast channels Admit-via-Unicast Since 0 < (t – t m-1 ) < (T – δ ) ≪ L, unicast channels are occupied for a much shorter duration compared to TVoD

9. 9 Performance Modeling Latency (average waiting time) Admit-via-Multicast Admit-via-Unicast Admission Threshold Channel Partitioning

10. 10 Waiting Times (1) Admit-via-Multicast w M ( δ ) = δ / 2 Admit-via-Unicast Arrival process λ u = ( 1 – δ / T R ) λ Service time Uniform distribution between 0 < s < T R – δ Approximation by Allen and Cunneen for G/G/m queue

11. 11 Waiting Times (2) Traffic intensity Coefficient of variation Average service time Server utilization Erlang-C function

12. 12 Admission Threshold

13. 13 Channel Partitioning Find the optimum number of multicast channel such that the resultant latency is minimized Theorem 1: The optimal proportion of available channels to multicast that minimizes the load at the unicast channels is given by

14. 14 Numerical Results Corresponding Latency Formula NVoD The latency is constant at 360(900)s for 10(20) movies TVoD

15. 15 Admission Threshold verus Queueing Delay

16. 16 Channel Partition versus Latency

17. 17 Latency Comparison With TVoD and NVoD

18. 18 System Capacity and Scalability (1) λarrival rate in customers/s ulatency constraint in seconds W UVoD latency fo UVoD W TVoD latency fo TVoD

19. 19 System Capacity and Scalability (2) 0.1 0.2

20. 20 System Capacity and Scalability (3)

21. 21 Simulation Results Environments Simulation program is developed in C++ using CNCL version 1.10 Run 31 days Model Validation Admission Rescheduling

22. 22 Model Validation (1)

23. 23 Model Validation (2)

24. 24 Admission Rescheduling (1) When Admission Rescheduling? For heavy system loads, a user by Admit- via-Unicast may waiting exceed the time to the next multicast of the requestd movie

25. 25 Admission Rescheduling (2)

26. 26 Interactive Controls (1) Using Unicast Channels Break current multicast video stream then restart at some point Treat interactive controls as new-video requests starting at the middle of a movie Could increase waiting for both and interactive requests

27. 27 Interactive Controls (2) Channel Hopping Client has a buffer large enough to cache T R s User pause at a movie time T p Case1: If resume before buffer overflow, nothing need to be done Case2: Once buffer is full, stop buffering Later resume immediately and determine the nearest multicast channel at movie time T m ≤ T p

28. 28 Conclusions This paper propose and analyzes an architecture that unifies the existing TVoD and NVoD Through admission-threshold and channel partitioning can achieve cost-performance tradeoff Results show large performance gain