Analysis of a method for improving perceptual quality of video in a multicast network

Examination of packet traces from the Internet reveals a significant number of multiple consecutive packet losses, which is detrimental to perceptual quality of Internet video. Capitalizing on the unordered-delivery property of Internet transport protocol (UDP) and the human visual system\´s insensitivity to the HF components of the video signals, a method has been proposed to attenuate the number of such burst losses by "transforming" them to singular isolated losses and to burst losses with fewer consecutive packets. With the use of an efficient layered coder the incoming video signals are segregated into base (essential) and enhancement (nonessential) data to be packetized in base and enhancement packets respectively. Using probability models, this paper presents analysis of the effectiveness of deliberately increasing the number of enhancement packets in order to protect critical video data from loss. Evaluation of the results from simulations as well as from probability models indicates that increasing the number of enhancement packets is indeed effective in reducing the number of base packets lost in a burst and that the number of isolated losses is increased