Statistical multiplexing gain for variable bit rate video codecs in ATM networks

Abstract

Asynchronous transfer mode (ATM) broadband networks will support variable bit rate video codecs, which are capable of maintaining a constant picture quality. To demonstrate this capability, a prototype hardware video coder has been developed in the Siemens Central Communications Laboratories. The prototype uses interframe coding, combined with a discrete cosine transform, and is able to reproduce the original picture quality, independent of signal sources or picture material used.A gain in transmission efficiency is expected when several video sources share a common ATM channel ('statistical multiplexing'). This paper reports on a series of measurements that have been performed using this coder for a large variety of video sources to determine the possible gain in transmission efficiency.The main results are: for realistic video phone scenes, up to about three times the number of signals can be transmitted compared to transmision with constant rate and the same picture quality, if the output signal of the coder has been smoothed over a period of one frame. Smoothing over shorter periods reduces the potential gain substantially. The statistical multiplexing gain increases with the duration of the picture sequences due to the criterion of constant picture quality. It varies very little with the acceptable packet loss rate.

On the efficiency of statistical—bitrate service for video

“…Yet the results show that SBR is not always yielding a lower delay than DBR. For instance, a call with 1.8 Mb/s sustainable rate and 8 Mb/s peak rate (a 5: I peak-ta-mean ratio is common for VBR video [6]) would require a declared maximum burst size of nearly 9 Mb in order to yield lower smoothing delay than a DBR connection with the same allocated capacity. This only assumes an average source rate that is 70 percent of the sustainable rate.…”

“…All studies of statistical multiplexing gains have assumed source characteristics that are known to a very high degree: for instance, they could be captured by a stochastic model with parameters fitted to real data [7], or when only the first two moments are used, they are still actual values for a real source [6]. What we have shown here is that a leaky-bucket description of a source does not provide enough infonnation about source characteristics to ensure a reasonable multiplexing gain in many cases.…”

“…This, in tum, means that it is better to reduce the bitrate by source coding and allowing at most a small amount of cell loss, compared to nearly lossless source coding and more cell loss in order to get more efficient multiplexing. For example, Heeke reports that the statistical multiplexing gain increases 20 percent for a video conference scene and 40 percent for a television scene when increasing the cell loss rate seven orders of magnitude from 10-9 to 10-2 [6]. An equally large increase in distortion caused by source coding would most likely allow a much higher reduction in needed transmission capacity for the signal.…”

Stochastic Features of VBR Video Traffic and Queueing Working Conditions: A Simulation Study Using Chaotic Map Generator

“…The first one is a videoconference session corresponding to three people in front of a 3-CCD camera, while the second is a long videophone session (a man carrying on a videophone call). These two sequences were produced by a prototype coder described in [8]. The last sequence is an example of entertainment video, namely the movie "Star Wars", corresponding to a frame by frame coding scheme similar to JPEG [9].…”

Fitting Long-Memory Models by Generalized Linear Regression

“…About 30 frames per second are processed. More technical details about this series are given by Heeke (1991) and Heyman, Tabatabai & Lakshman (1992). The periodogram in log-log coordinates is displayed in Fig.…”