So What About MPEG-4, Anyway?
Recently, in some industry newsletter, someone suggested that the "spectrum crisis" we are purported to be experiencing (i.e., television stations are sitting on spectrum that is urgently required to expand broadband services), might be alleviated by the use of MPEG-4, rather than MPEG-2, in U.S. DTV broadcasting. What is MPEG-4, anyway?
A LOOK BACK
In the United States, digital television broadcasting, as codified in the ATSC standards, employed the latest, state-of-the-art video coding technology available at the time the standards were established: MPEG-2, Part 2.
This international standard video coding technology was jointly developed by the ISO/IEC Motion Picture Experts Group (MPEG), and the ITU Video Coding Experts Group (VCEG), and is known by two titles: ISO/IEC 13818-2 and ITU-T H.262.
It was also adopted by the European Digital Video Broadcasting (DVB) consortium for DVB Digital Television Terrestrial Broadcasting (DTTB), which, at the outset, only broadcast SD, not HDTV.
MPEG-2 was proposed for ATSC standardization in the United States by the "Grand Alliance" of DTV broadcasting proponents. It had the significant advantages over the other system proponents' compression schemes that it was 1) an international standard, not a proprietary compression format; and 2) that it was highly versatile, unlike the dedicated HDTV coding schemes of the other proponents.
MPEG-2 was initially developed for SD compression. It is sometimes described as a "tool kit;" users can avail themselves of whichever tools in the kit they find useful, leaving others behind. It was also developed under the overarching concept that while the encoder modules could be complex and costly, the decoder modules should be simple and inexpensive.
The predecessors to MPEG-2 were ITU-T H.261 and MPEG-1. ITU-T H.261 was a teleconferencing compression system. It introduced the concept of the macroblock, which is a part of MPEG-2 as well. It also established the concept of specifying in the standard only how to decode the compressed signal, leaving it up to the designer to develop the coding functionality.
MPEG-1, standardized as ISO/IEC 11172, was based on H.261, and was a compression system optimized for the compression of relatively low bit-rate video. It also introduced the MP-3 audio compression format.
The Harmonic Electra 8000 encoder supports HD and SD compression, in MPEG -4 AVC and MPEG -2.
MPEG-2, Part 2, is formally known as ISO/IEC 13818-2 and also as ITU-T H.262. It is built on the basis of MPEG-1 and H.261, with the difference that it includes tools for compressing interlaced pictures as well as progressively scanned pictures.
It employs the concepts of macroblocks and discrete cosine transform (DCT) encoding. It has proven highly versatile, in that it is used for both SD and HD video. It is what we colloquially refer to as "MPEG-2."
MOVING ON TO MPEG-4
MPEG-2 has served us pretty well in ATSC broadcasting, but time marches on and technologies improve. Today, we have the tools incorporated into MPEG-4, Part 10, which is some 10 years younger than MPEG-2, available to us. The standard, like MPEG-2, is jointly maintained by the ISO/IEC and ITU-T, and is formally known as ISO/IEC 14496-10 – MPEG-4, Part 10, and ITU-T H.264.
Interestingly, H.264 is freely available on the ITU website, but fair warning, it is 676 pages long! MPEG-4, Part 10 is built on its predecessors, and, like them, is block-based, using DCT coding, and Anchor (I) Frames, Predictive (P) Frames, and Bidirectional (B) Frames.
Its chief advantage over MPEG-2 is that it is considerably more efficient in its coding. Video of comparable quality can be compressed to a considerably lower bit rate, often claimed to be half the bit rate, of MPEG-2.
Today, the desire to have multiple subchannels on a broadcast signal, along with the imminent mobile ATSC implementations, is generating a sort of spectrum crisis within a single 6 MHz television channel. It would certainly be nice to be able to pack twice the content into the channel, and it might make it easier to conceive of a reduction in the number of 6 MHz channels occupied by broadcast television channels.
There is a big problem to be overcome, before this can happen. That is, what to do about the now very large installed base of DTV receivers that decode MPEG-2, but have no idea what to do with MPEG-4.
As mentioned, European broadcasters did not broadcast HDTV when they initiated their DTV broadcasting efforts. When they did launch terrestrial HDTV broadcasting, there was no installed base of DVB MPEG-2 HDTV receivers. This gave them a tabula rasa to work on, and they decided to use MPEG-4 for HDTV service.
There is no question that using MPEG-4 rather than MPEG-2 would result in considerably improved spectral efficiency. How to get there, however, is a very big question. Perhaps a new technology might be devised that could somehow bridge the gap, and avoid obsoleting the entire installed base of ATSC receivers in the field. It is likely that work is proceeding behind the scenes on doing just this. We can hope.
Randy Hoffner is a veteran of the big three TV networks. He can be reached through TV Technology.