Next: Traffic Prediction
Up: Motivations
Previous: Impact of the Quality-Affecting
  Contents
  Index
A third important problem in traffic engineering is rate
control. Traditionally, rate control protocols are designed based on
passive network measurements (loss, delay, bandwidth,
etc.) [26,41,44,77,107,108,124,126].
However, human subjects are the final end-users of real-time multimedia
applications. The direct reaction of these end-users is often neglected
in the design of these protocols, because it has been difficult to
quantify in real time their perception in a way allowing to take into
account the direct effect of both network and codecs's parameters. The
solutions to the previously mentioned problems that we present in this
thesis (namely the real-time quality assessment of real time
multimedia transmitted over packet networks and the understanding of
the impact of the quality-affecting parameters on user's perception)
may make it possible to think in designing new kinds of ones. These
protocols, in addition to the passive networks measurements employed
in the traditional protocols, could take into consideration the direct
influence of the end-users' perception in real time. We concentrate
in this dissertation on providing a new rate control protocol that
integrates both network measurements and users' perception. Among the
benefits of having such a protocol, let us mention the following:
- We can maximize or guarantee the delivery of the best possible multimedia quality given the network state.
- It helps in reducing network congestion.
- It keeps the stream in a TCP-friendliness state and hence does not contribute to collapse the network.
- Based on the rate controller and the real-time multimedia quality assessment, multimedia codecs could be re-engineered to compress the original signal based on users' perception and the network state rather than using only the simplistic SNR or PSNR objective quality measures.
It is important to underline that the design of these protocols was not
possible without having a tool that measures or evaluates in real time
the user's perception of real-time multimedia quality given the parameters' values. In this dissertation, a third objective of our work is to propose
such a protocols.
Next: Traffic Prediction
Up: Motivations
Previous: Impact of the Quality-Affecting
  Contents
  Index
Samir Mohamed
2003-01-08