A feedback based rate control algorithm for multicast transmitted video conferencing

  • Authors:
  • Hugh M. Smith;Matt W. Mutka;Diane T. Rover

  • Affiliations:
  • -;-;-

  • Venue:
  • Journal of High Speed Networks - Special issue on multimedia networking
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a feedback based rate control algorithm forvideo conferencing on the Internet. This algorithm adaptivelycontrols the transmission rate of a video stream, which is beingmulticast to a heterogeneous group of receivers. The fact that thesender is transmitting a single video stream to many receiversintroduces a number of issues. These issues include the variablenature of the available network resources and the fact that thereis no easy way to determine the availability of these resources. Inaddition, even if we were able to track closely the changingnetwork conditions, the best transmission rate for the aggregatemeans a trade off between high and low end receivers.The algorithm presented in this paper, called the TargetBandwidth Rate Control Algorithm, dynamically controls the outputrate of the video coder by receiving and processing bandwidthlevels from each receiver in the videoconference. The goal of thisalgorithm is to maximize the aggregate amount of video displayableat the receivers. This algorithm also allows us to limit the worstcase loss experienced by the low-end receivers. In addition toconsidering the available network bandwidth in the feedbackinformation, this algorithm allows for workstation and userrequirements to be considered when determining the senders outputrate.In order to show the effectiveness of this approach, we firstanalyze the algorithms performance in a simulation environment. Inthis set of tests we study the effects of different configurationsof receivers in terms of the number of receivers and availabilityof resources. Additionally, we have implemented this algorithm inthe VIC video conferencing system in order to analyze itseffectiveness under real network conditions.