We describe a mechanism for scalable control of multicast continuous media streams. The mechanism uses a novel probing mechanism to solicit feedback information in a scalable manner and to estimate the number of receivers. In addition, it separates the congestion signal from the congestion control algorithm, so as to cope with heterogeneous networks. This mechanism has been implemented in the IVS videoconference system using options within RTP to elicit information about the quality of the video delivered to the receivers. The H.261 coder of IVS then uses this information to adjust its output rate, the goal being to maximize the perceptual quality of the image received at the destinations while minimizing the bandwidth used by the video transmission. We find that our prototype control mechanism is well suited to the Internet environment. Furthermore, it prevents video sources from creating congestion in the Internet. Experiments are underway to investigate how the scalable probing mech...

he video delivered to some of participants will be of low quality (since packet losses and/or delays will be high on some branches of the tree). The second case is safest, but it results in disturbing the bulk of the participants. In practice, it is not clear how to choose an adequate value for this fraction. In IVS, we set the fraction to a few percent [2, 3]. Ideally, however, the source should be able to single out the parts of the multicast tree that experience congestion. In order not to disturb the bulk of participants in the tree, these branches should be treated separately. Two possible solutions include video gateways, and using some form of layered coding. Video gateways or layered coding schemes are not new ideas. Our contribution is to identify and discuss the issues associated with using these techniques in the Internet. We illustrate our points with the H.261[10] software coder of IVS. 2 Video gateways Video gateways take

There exist a number of algorithms for compressing digital video, such as the ISDN H.261 standard, and MPEG I and II. With all these algorithms, it is possible to vary the degree of compression within the codec and thus the output data rate, by manipulating the parameters of the algorithm. Using intelligent codec implementations, it becomes possible to send compressed video over packet switched networks and adapt the rates at which the compressed video is produced to maximise video quality, whilst protecting the network from congestion. This thesis presents two viable schemes which use results from human factors experiments that prove this is possible with minimal support from the intermediate switches. First we derive the requirements for transmitting video over packet switched networks, surveying the reported work on human factors, and survey and compare the various algorithms for compressing video and determine their suitability for use over packet switched networks. We produce a t...