Only by ensuring high levels of all three, can IPTV service providers begin to capture new subscribers and keep them loyal. Because MPEG-2 and MPEG-4 AVC are similar technologies, this report includes information for both MPEG-2 and MPEG-4 AVC operations. However, the emphasis in this report is on MPEG-4 AVC, since most (if not all) new IPTV systems are forecasted to use MEG-4 AVC from 2007 forward.

1.2 Situation

Telcos are fighting an uphill and increasingly fierce battle against incumbent competitors, so VQ/V-QoS and QoE are key considerations. To deliver video services, they must not only engineer their networks more deliberately and monitor them more carefully than ever before, they also must quickly come up to speed with unfamiliar IPTV infrastructure, such as the headend, TV-enabling software and CPE.

Complicating matters is the fact that many crucial IPTV infrastructure and service-enabling elements are immature and have yet to be proven in scaled production deployments, which creates the risk of unexpected problems that affect VQ and QoE but don't necessarily have quick fixes. We specifically make reference to IPTV middleware and interactive television applications.

1.3 IPTV Observations Relating to Quality

Because the IPTV ecosystem is complex, this report breaks it up into seven domains, which are identified in Section 3 and discussed in detail in Section 4. The complexity of this ecosystem makes it very expensive to implement. To minimize capital equipment and software costs at deployment, IPTV operators assume an oversubscription model - deploying enough equipment to serve under average conditions with statistically-determined peaks, as opposed to being designed for full-time peak capacity.

The underprovisioning associated with the oversubscription model forces operators to balance video quality against costly and scarce infrastructure resources while striving to attract consumers from competing options and keep them loyal. IPTV delivery has inherent sensitive dependencies between video encoding (compression), the bit-rates available for the delivery of aggregate and individual streams, the method of video transport, and the decoding of video.

Delivery errors attributable to the network are responsible for a majority of V-QoS and QoE issues. Unlike data, V-QoS parameters must go beyond the traditional datacomm parameters such as cyclical redundancy checking, to accommodate long sessions and a mix of different types of content.

An MPEG-2 video stream is a hierarchically structured sequence of pictures (see Figure 1-2). The sequence is composed of several pictures. Each picture is in turn composed of slices, which are series of macroblocks.

Each macroblock (16 x 16 pixels) contains four blocks (4 x 4 pixels) of luminance samples and two, four or eight blocks of chrominance samples, depending on the chrominance format.

The MPEG-4 standard was also developed by the MPEG organization, and is similar to MPEG-2 in that it also uses I-, B- and P-Frames to form GoPs that are packetized and encapsulated for delivery over an IP network.

Finally, the architectural standards, test metrics and implementation best practices that impact IPTV video quality are either in the process of being adapted from non-IPTV digital television standards; or are still in the proposal stage and therefore still fluid; or they are lacking altogether. Fortunately, MPEG video content reflects well-defined standards for the structure and packaging of the digital video content itself.

However, operators should not be lulled into having a false sense of security by the current standards and proposals. Metrics for network-related QoS errors such as jitter and packet loss, and for QoE issues such as latency, are much more stringent for IPTV than many of the existing standards allow. Some existing standards, in fact, were established for IP voice, not video, and don't take TV basics like channel change into account.

This report is part of the IPTV Tracking Service published by the Multimedia Research Group, Inc.