Paper review : Quality Adaptation for Congestion Controlled Video Playback over the Internet (RHE99)

• Goal
  To provide an AIMD based unicast congestion control machanism for the streaming audio and video applications.

• Contribution
  This paper presents an efficient scheme that dynamically adjusts the buffer distribution among the active layers as the available bandwidth changes. This scheme also allows the server to trade short-term improvement for long-term smoothing of quality by introducing a smoothing parameter.
  

• Main ideas
  • For multimedia applications, frequent variations in the transmission affects the quality severely, and the applications should provide a end-to-end machanism to solve this problem.
  • Layer adding and dropping provides more flexibility to layered encoding.
  • The server may use a smoothing parameter to trade the long-term and short-term qualities.

• Evaluation
  • Significance rating: 3
    Streaming audio and video applications are becoming increasingly popular on the Internet. The machanism presented in this paper give a good investigation on how the applications can overcome the bandwidth fluctuations on the Internet by adopting an end-to-end machanism.
    

  • Convincing rating
    The author use the analysis and simulation methods to prove the efficency of the machnism, but seems the authors didn't try it on the Internet. It seems like layer addition and dropping is defined on each specific subflow, so do we need to know the exact variation for each subflow? If so, is the machanism still scalable?

  • Limitation
    This work is for the unicast congestion control instead of the multicast context. It is based on the AIMD algorithm instead of a more general congestion control context. It also needs the applications divided the multimedia into several layers in advance.

• Conclusion
  Much topics are available in the field of the quality of service.

  10/17/01