Open-Loop Algorithms - Scalable Continuous Media Streaming Systems - page 329

First, larger values of N result in more frequent channel switching, and as discussed earlier
in Section 18.7.1, there is some delay from the time the client leaves a multicast group to the
time the network switch stops forwarding the multicast data. This results in some duplicated
data being transmitted to the client, only to be discarded by the client's operating system. The
second reason is due to the specific network switch we used in the experiment. Our results show
that there seems to be bugs in the switch's hardware, resulting in some random multicast data
transmitted to the client after the switch has pruned the multicast tree. This specific problem
is easy to miss because the random multicast data will be discarded by the client's operating
system (as the client has left the multicast group already) and thus will not cause any data
transmission or application error. We expect this problem to be resolved in future revisions of
the switch hardware.
18.8 Summary
In this chapter we have reviewed some open-loop multicast streaming algorithms in the con-
text of a taxonomy, which classifies the algorithms according to the media segmentation and
transmission bandwidth schemes adopted. To further illustrate the design and trade-offs in de-
veloping an open-loop algorithm, we described in detail as well as analyzed the performance
of the Consonant Broadcasting algorithm. We also addressed some practical issues in the im-
plementation and deployment of the Consonant Broadcasting algorithm, which are likely to
be applicable to other open-loop algorithms as well.
Unlike the closed-loop algorithms, the resources consumed by open-loop algorithms are
fixed irrespective of the system load, i.e., number of concurrent users. The upside is that open-
loop algorithms will be very cost-effective in serving popular media streams (e.g., popular
movies) of a large user population. The downside, however, is that for unpopular media streams
the resources requirement could exceed those of closed-loop algorithms, which are more
efficient when the system load is lighter. In the next chapter, we illustrate a hybrid approach
to multicast streaming, combining elements of both open-loop and closed-loop algorithms in
the same architecture.
References
[1] S. Viswanathan and T. Imielinski, Metropolitan Area Video-on-Demand Service Using Pyramid Broadcasting,
IEEE Multimedia Systems , vol. 4, 1996, pp. 197-208.
[2] K.A. Hua and S. Sheu, Skyscraper Broadcasting: A New Broadcasting Scheme for Metropolitan Video-on-
Demand Systems, Proceedings of the ACM SIG-COMM '97 , Cannes, France, Sept. 1997, pp. 89-100.
[3] L. Gao, J. Kurose, and D. Towsley, Efficient Schemes for Broadcasting Popular Videos, Proceedings of the 8th
International Workshop on Network and Operating Systems Support for Digital Audio and Video , Cambridge,
UK, July 1998.
[4] L.S. Juhn and L.M. Tseng, Staircase Data Broadcasting and Receiving Scheme for Hot Video Service, IEEE
Transactions on Consumer Electronics , vol. 43, no. 4, Nov. 1997, pp. 1110-1117.
[5] L.S. Juhn and L.M. Tseng, Harmonic Broadcasting for Video-on-Demand Service, IEEE Transactions on Broad-
casting , vol. 43, no. 3, Sept. 1997, pp. 268-271.
[6] J.F. Paris, S.W. Carter, and D.D.E. Long, A Low Bandwidth Broadcasting Protocol for Video on Demand,
Proceedings of the 7th International Conference on Computer Communications and Networks , Lafayette, LA,
USA, Oct. 1998, pp. 690-697.