Donald Bren School of Information and Computer Sciences

The transport of high quality live video requires a video service that does not rely heavily upon accurate a priori traffic parameter estimation (as VBR does) and does not unnecessarily sacrifice multiplexing gain and visual quality (as CBR does). A new service capable of supporting the transport of video on a guaranteed as well as a best effort basis is introduced. This service, called quasi-VBR service, is based on the ATM Forum's EPRCA rate control algorithm [4], a modified version of Sriram's T_1/T_2 service discipline [7], and an adaptive video encoding scheme. The results of a simulation study in which quasi-VBR video and ABR data traffic interact are presented. Actual MPEG-1 video sequences are used. Results show that quasi-VBR video service is capable of rapidly adapting to changes in network conditions, providing high quality video service and excellent network resource utilization.