HDTV originated in Japan, where the original HDTV scanning format, 1035i, was developed. A pioneer of HDTV, Japanese public broadcaster NHK has been an aggressive and vocal promoter of interlaced HDTV, to the point that Japanese broadcast equipment manufacturers have until recently demonstrated reluctance to publicly embrace progressively scanned DTV formats. Privately owned Japanese broadcaster NTV has been active in the development of progressive formats, but has not yet announced which formats it will support.
Some recent field-testing of BST-OFDM (bandwidth segmented orthogonal frequency division multiplexing) in Tokyo produced results that have led the Radio Regulatory Council to add 720p HD the table of authorized scanning formats for terrestrial DTV broadcasting in Japan.
The digital television field tests were conducted by an industry organization, the Digital Terrestrial Broadcasting Tokyo Pilot Project, and NTV. This test marked the first time 720p has been broadcast using OFDM.
We know from previous columns that OFDM may employ any of a number of modulation densities, guard intervals and forward error-correction strengths. We saw that COFDM as specified in the European DVB-T standard, for example, may be modulated with several densities up to 64 QAM, and employ one of a number of guard interval durations and forward error-correction strengths.
We also know that there is no such thing as a free lunch - and in the case of OFDM, there are direct trade-offs between modulation level, guard interval and error-correction parameters on the one hand and payload data rate on the other.
ONE BETTER
The BST-OFDM system proposed for Japan goes COFDM one better by exploiting the fact that some OFDM carriers may be modulated differently from others within the same multiplex. The 6 MHz television channel may therefore be "segmented," with different segments being modulated differently, and used for different services.
It is possible, for example, to send an audio service on a segment that includes a segment comprised of some number of carriers, a data service on another segment, and a television service on yet another segment - all within the same 6 MHz television channel. Further, these may be modulated with different parameters so that, for example, the audio and data services could be optimized for mobile reception, while the television service is optimized for stationary reception in a high-multipath environment.
The result reported from the tests is that in order to transmit 1080i HD while maintaining the original picture quality required 22 Mbps, which is very close to the payload limit within a single television channel, using the highest order modulation, the shortest guard bands and the weakest forward error correction that may be applied. 720p transmission, on the other hand, may be compressed to about 15 Mbps without picture degradation, leaving adequate payload for another SD television service or a number of audio and/or data services within the same channel.
The Tokyo tests have confirmed the fact that a progressive format may be more efficiently compressed than an interlaced format. This can be seen when we look at the numbers. 1920 x 1080i/30 fps in Y, R-Y, B-Y format at 10 bits has a total payload data rate of about 1.24 Mbps, while 1280 x 720p/60 fps at 10 bits has a total payload data rate of about 1.11 Mbps. The uncompressed data rates are very close to the same, yet the 720p/60 signal was successfully compressed to a data rate 7 Mbps lower than could be achieved with the 1080i/30 signal.
MORE SERVICE PER SPECTRUM
This is really more than a matter of packing more services into the same amount of spectrum. If Japanese broadcasters wish to take full advantage of the ghost-busting capabilities of OFDM, and to broadcast HDTV, they must use more than minimal guard bands and forward error correction. Otherwise, the signals will be very fragile - and certainly reception of them using indoor antennas will be difficult if not impossible, given the heavily urbanized nature of Japan.
If very short guard intervals are used, there is no way to protect against longer reflections, because any reflection longer than the guard interval will not be ignored by the COFDM demodulator. Further, the more data packets that are "missed" because of inadequate guard intervals, man-made noise, or for any other reason, the more demand is placed on the forward error-correction scheme - and a weak forward error-correction scheme does not have the headroom to cope with a large percentage of missed packets.
Those nasty laws of physics rear their ugly heads again! They may even convince Japanese broadcasters of the benefits of progressive scan! Can Europe be far behind?
