From specialized image-compression ICs to op amps, manufacturers have touted high-definition television (HDTV) as a potential application for products introduced within the last few years. This promotion occurred despite the fact that manufacturers had no firm notion of the actual form that US HDTV would take. However, that form is finally beginning to take a very definite shape.

Since 1987, when the FCC began organized efforts to draft a broadcast standard, US HDTV has gone from a system with digital compression and analog transmission to a hybrid digital/analog transmission system to the current all-digital system.

The proposed US system places a heavy emphasis on computer-compatible progressive-scanning techniques—as opposed to traditional NTSC TV's interlaced mode—and MPEG-2 compression and decompression techniques. According to Glenn Reitmeier, the director of the High Definition Imaging and Computing Laboratory at the David Sarnoff Research Center in Princeton, NJ, this emphasis points to a future in which MPEG may become the de facto standard for the multimedia industry. "We're on our way to a very interoperable format between computers and HDTV," says Reitmeier. "Future consumer products may have a much more multimedia feel than does traditional TV," he adds.

The proposed system, which its proponents describe in terms of layers (Fig 1), is a very flexible system that encompasses multiple picture formats and frame rates and a flexible transport channel that shares video and audio signals.

The current activity on the US digital HDTV standard—for which some key technology decisions were announced last fall and others are due early this year—is an effort of both compromise and expediency. After the 7-year process of proposals, testing, and FCC recommendations, the surviving companies—and former opponents—banded together last year to form the Grand Alliance. Working together, the members hope to bring HDTV signals and sets into US homes as early as 1996.

This Grand Alliance includes AT&T, General Instrument Corp, the Massachusetts Institute of Technology, Philips Consumer Electronics, Thomson Consumer Electronics, the David Sarnoff Research Center, and Zenith Electronics Corp (see box, "For more information... ").

According to its members, the alliance could save a year or more in HDTV implementation by reducing the risk of inconclusive test results and the possibility of legal challenges. All members hope that by the end of this year or early next year, the FCC advisory committee will make its final and complete HDTV recommendation. At that time, you can expect a flurry of design activity to begin. Each member is currently designing or actively building pieces of the prototype for testing and evaluation (see box, "Looking ahead").

Although it would be premature to start a full-scale product development before the FCC approves the final standard, the alliance has defined the basic functional blocks (Fig 2) and specified some minimum system requirements. Final HDTV products will be very digital and processor intensive.

Last October, the alliance decided on four main technologies that will be at the heart of the digital HDTV system: digital video-compression technology based on MPEG-2 parameters, including the use of B-frames (bidirectional frames for motion compensation); a data-transport system based on packets of virtually any combination of video, audio, and data; interlaced- and noninterlaced- (progressive) scanning capabilities with a heavy emphasis on progressive; and the 5.1-channel Dolby AC-3 audio technology for digital surround sound. However, at the time, the alliance did not make one important decision: which transmission scheme the HDTV system will use (see box, "Looking ahead").

The alliance also decided on the following scanning formats: 24-, 30-, and 60-frame/sec progressive scan with a pixel-by-line format of 1280×720 and 24- and 30-frame/sec progressive scan with a format of 1920×1080. The system will also perform a 60-frame/sec interlaced scan with a format of 1920×1080. These formats provide a foundation for the migration to the ultimate goal of a 60-frame/sec, 1920×1080 progressive format as soon as technically feasible.

MPEG-2 plays a major role

The chosen digital video-compression technology, based on MPEG-2 parameters, forms a major part of the evolving standard. MPEG-2 is not one standard but a kind of tool kit of syntactic elements that encompasses a range of compression grades that vary in performance and cost. Fig 3 shows the elements of this toolbox—referred to as profiles—vs the formats, or levels, on the y-axis. The profile refers to one of the four types of compression: simple, main, main+, and next.

A given decoder can work at its own profile and its own or lower level. A decoder with a simple profile uses only forward-motion prediction. A main profile implies the use of bidirectional prediction, which requires two frames of storage but improves picture quality. Thus, conforming to a main profile implies a receiver with much more memory than one that conforms only to the simple profile. Operating at different levels requires vastly different data rates, as Fig 3 shows.

Although the level and profile syntax defines a certain level of performance, MPEG-2 does not specify any details of the hardware and software architectures that produce this performance. However, the syntax does imply many things. A decoder that performs at the MPEG-2 main profile and high level implies how fast the decoder must operate and how much memory it needs to have. A system that can perform to multiple MPEG formats requires some electronic format-conversion circuitry. System-performance specifications to consider include speed, I/O bandwidth, and memory size. The alliance has yet to nail down the numbers that correspond to these specifications. However, system clock speeds may be as high as 75 MHz.

The MPEG-2 levels and profiles provide the flexibility necessary to deal with compressing, transmitting, and decompressing types of pictures. The need to transmit data at megabits/second over the FCC-mandated 6-MHz channel requires some form of lossy compression. The optimum compression scheme depends on a picture's contents and requires a compromise between spatial resolution, frame rate, and amount of acceptable compression artifacts. Optimum resolutions and rates vary for different kinds of pictures.

With the standard close to completion, much product development will soon be necessary, particularly in the area of displays and powerful compression and decompression ICs. MPEG-2 ICs are already starting to appear. For example, AT&T Microelectronics and SGS-Thomson Microelectronics are both sampling MPEG-2 ICs. The first of a family of MPEG-2 devices planned by AT&T, the AC6101 decodes all the MPEG-2 video layers in real time without external processor support. The device uses a 27-MHz clock and just 8 Mbits of external memory. The chip will be available in volume in the first half of this year. The company hasn't set firm pricing but expects the IC to cost between $50 and $75 in large volumes.

SGS-Thomson's Sti3500 decodes MPEG-1 and -2 bit streams in real time up to a 16-Mbps input-data rate. Samples are available now, and production quantities will be available this quarter. Price is $48 (100,000).

In addition to decoder ICs, the HDTV standard will impact the design of many other systems, including receivers, displays, studio and transmission equipment, peripheral equipment, programming and software-development systems, and semi-conductors. Within a year, specific details will be available so that designers can begin work on these products.