1. Constrained Consonant Broadcasting- A Generalized Periodic Broadcasting Scheme for Large Scale Video Streaming W. C. Liu and Jack Y. B. Lee Department of Information Engineering The Chinese University of Hong Kong ICME 2003

2. Outline Poly-harmonic Broadcasting Constrained-Consonant Broadcasting Performance Evaluation

3. Objective CCB can be considered as a generalization of the Poly-harmonic Broadcasting scheme incorporating two important constraints, namely client access bandwidth and client buffer requirements.

4. Notations

5. L B S1S1 S1S1 S1S1 S1S1 S2S2 S3S3 SNSN

6. Harmonic Broadcasting (HB) Divide a video into N equally-sized segments Each segment S i, for 1 ≤ i ≤ N, is broadcast repeatedly on its own channel with a bandwidth ( b/i ) HB does not always deliver all data on time

7. Harmonic Broadcasting –The i th segment of the movie S i is equally divided into i sub-segment( s ) { S i, 1, S i, 2 --- S i, i } –Let the i sub-segment(s) of S i be put on a logical channel C i, the bandwidth of C i is b / i.

8. Harmonic Broadcasting The total bandwidth(B) allocated for the movie is as follows: Where H N is called the harmonic number of N B = b + b/2 + b/3 + b/4 = 2.083b H N = 1 + 1/2 + 1/3 + 1/4 = 2.083

9. An illustration of the first three streams for a video under harmonic broadcasting Play Rate : b Receive Rate : b/2

10. Poly-harmonic Broadcasting Two major changes: 1) The client STB starts downloading data from the moment a customer requests a specific video instead of waiting until the customer begins watching the beginning of the first segment. 2) Fixed wait policy.

11. Poly-harmonic Broadcasting Divide the video into N equal segments of duration U =( L / N ) Segment Si at a transmission rate –B i = b / ( m + i - 1) No client can start consuming the S 1 of the video before having downloaded data from all N streams during a time interval of duration T = mU, m is an integer ≧ 1. Segment S i will not be consumed until ( m + i –1) U time units have elapsed.

12. Poly-harmonic Broadcasting ( m =2)

13. Poly-harmonic Broadcasting The Bandwidth –If N = k * m Since T = mU; U= L/N; T = L/k ; k 愈大 則 waiting time 愈小

14. Poly-harmonic Broadcasting (PHB) Unlike the original HB, the Poly-harmonic Broadcasting scheme guarantees continuous video playback and at the same time can achieve near- optimal performance. Provides the same maximum waiting time as harmonic broadcasting protocol while consuming significantly less bandwidth.

15. Poly-harmonic Broadcasting Poly-harmonic Broadcasting requires a client to be able to receive all broadcasting channels simultaneously and has a buffer large enough to store up to 37% of the whole video.  difficult to implement!

16. Constrained-Consonant Broadcasting (CCB) Divide the video into N equal segments of duration U =( L / N ) Target latency T = mL/N=mU Classify broadcasting channels into two types, namely Type-I and Type-II channels.

17. Type-I Channels Channels are allocated with progressively less bandwidth as given by for the i th channel, where n 1 is the total number of Type-I channels. For Type-I channels, the client is required to start receiving video segments upon entering the system and begin video playback in T seconds. For m =2, B0 = b/2, B1 = b/3, B2 = b /4…

18. Type-I Channels Solve for n 1, such that the following constraints are satisfied: if we remove both the client access bandwidth and client buffer constraints  n 1 = N, CCB  PHB This PHB can be considered as a special case of CCB without client access bandwidth and client buffer constraints. (a) (b)

19. Type-II Channels Type-II channels are divided into groups of consecutive channels. Once a client completes receiving a video segment, the corresponding channel will be released ---begin receiving a new group of Type-II channels.

20. Type-II Channels Let n 2,j be the number of channels in group j, of which is created after channel j is released, where j =0,1,…, etc. Then the bandwidth allocation for channels in group j is given by BjBj BiBi

21. Type-II Channels Solve for n 2, such that the following constraints are satisfied: (c) (d) and where

22. Performance Evaluation