<title>Networked video: MPEG transport, ATM, or IP?</title>

Abstract

Translate Article

ABSTRACT Soon the days of separate transmission facilities for video will be over, and video will be just another traffic type in amultipurpose network. These networks today employ a variety of transport mechanisms, for historical as well as technicalreasons, and, at least experimentally, enterprises, network operators, and service providers are sending video over thesenetworks with their associated transport mechanisms. But video is not just a random bit stream, so it is natural to ask howwell the transport mechanisms perform when transporting video. In this paper we examine what video needs from capture toplayback and how these can be provided by the end system or the network using the MPEG transport stream, asynchronoustransfer mode, and the Internet Protocol. We concentrate on the Internet Protocol because of the opportunity for networkedvideo applications that its pervasion provides.Keywords: IP Video, IP Multicast, IP Premium Service, MPEG, ATM Video 1. INTRODUCTION As video becomes increasingly represented in digital form, it enters the networked world, in which it shares network facilitieswith other types oftraffic. It does so at a time of upheaval and transition in networking, on severn! fronts. High-bandwidthdigital networks reach more and more homes, and most of these employ cable television facilities. For these, the MPEGtransport stream carries MPEG-2 video along with its associated audio channel and also data. Digital subscriber lines, whilenot as prevalent as digital CATV networks, allow a few megabits of traffic (downstream anyway) and most commonly relyon ATM for basic transport. In the Internet and most enterprise networks, the Internet Protocol (IP)' dominates, and legionsof engineers worldwide strive to make IP the foundation for all data, voice, and video traffic.New advances in [P networking allow high quality video services for both continuous playback and interactive applications.The IP technologies facilitating support of high quality video on IP networks are: 1) dynamic algorithms used by videoterminals for smoothing jitter and clock slippage; 2) extended IP transport services providing reduced jitter, and 3) IPmulticast services for distributing broadcast video to large populations. This paper outlines a model for video transmissionover IP then contrasts the service to video transmissions over MPEG transport and ATM transport. We also consider theimportant role ofthe end system in augmenting what the network can do.1.1 Differences in VideoTo make sure we do not compare apples to oranges, we must keep in mind that there are different forms ofvideo, and this hasa bearing on the network transport. We assume that entertainment video is coded natively using the MPEG-2 programstream, requires 1 .5-6 Mb/s for sufficient quality, lasts from several minutes to several hours, conforms to a standard imagesize, and is used in urndfrectional applications rather than interactive ones; thus delay jitter is more important than absolutedelay through a network. Moreover, viewers typically pay to receive entertainment video and so demand high quality.Conferencing video is coded using the H.32x family of standards, requires from around 30 kb/s to over 1 Mb/s, takes ondifferent image sizes and aspect ratios, and because of its interactivity suffers in usefulness from absolute delay as well asdelayjitter. Quality is important but the viewer typically does not pay to receive the video and is therefore less demanding ofquality; also, the audio quality is usually more important because of the nature of the communication between live people.Web streaming