Joint Source-Channel Coding With Unequal Protection For Real-Time Scalable Video Transmission

Abstract

In real-time video transmission, the distortion of video content may be caused by video compression in source coding and packets loss in transmission. Meanwhile the source coding distortion and transmission distortion are often mutually affected. There is a fundamental tradeoff between them for achieving the best video transmission quality. In order to make wise use of network resources, this paper proposes a video transmission scheme in which the source coding distortion and transmission distortion to the video quality are formulated and fountain code is used as the channel coding for unequal protection in scalable video transmission. The network packet loss and the characteristic of scalable video are quantitated in our proposed rate distortion model, which formulates the real-time video transmission system in a bandwidth limited network. We introduce a computerized algorithm. The computerized algorithm will determine the optimal source rate of the video coding and the optimal redundancy amount assigned to each scalable video layer. Our proposed algorithm can guarantee that the optimal quality of real-time video transmission will be achieved and the total real-time video transmission distortion will be reduced dramatically.