Layered Multicast with Multiple Multi-Layered Video Sources and Multiple Shared Channels in Heterogeneous Environments
万俊伟,卢锡城
DOI: https://doi.org/10.3321/j.issn:0254-4164.2001.03.015
2001-01-01
Chinese Journal of Computers
Abstract:Multicast distribution of video is an important component of manyexisting and future networked services, such as video conferencing, distance learning, remote presentation and media-on-demand, and improvements in network delivery infrastructure and increases in end-system processing power have made these applications feasible. The main problem complicating multicast video transport is heterogeneity and variation of network transmission resources, which make it extremely difficult to agree on acceptable traffic characteristics among multiple receivers of the video stream. An attractive solution to this problem is to use an adaptive, multi-layered video transmission scheme, in which each source generates a layered media stream that is striped across multiple network channels. In multipoint-to-multipoint video multicast applications such as video conferencing, channels are always shared by all potential senders, any sender can send its video layers in any shared channels. In this multipoint-to-multipoint, shared channels, and layered video multicast model, one key problem is how to determine the mappings from video source layers onto shared multicast channels dynamically, which affects directly the whole video receiving quality of the session and the bandwidth utilization of the network. Classic approach is sequential mapping. It treats each sender equally, but introduces a bandwidth scaling problem at each shared network channel as the number of sources increases. Moreover, it can not adapt to the variation of network transmission resources and session status. In this paper, we design an adaptive layer mapping algorithm that relies on feedback from the receivers. The receivers feed back information about their current interested sender and receiving rate to one control node periodically, and the control node adjusts mapping strategy according to current feedback information dynamically. It is proved that our algorithm always achieves higher total video receiving quality than the sequential layer mapping algorithm with high bandwidth utilization and little complexity.