The design of the old DAB system

The Eureka 147 project that led to DAB being designed was launched in 1986 following promising results in the field of audio compression, and over the next few years the component technologies that make up DAB were chosen.

One of the earliest technologies chosen was the OFDM modulation, which has turned out to be the one and only good decision made in the design of the old DAB system.

Then in July 1990, the MPEG Audio Group set up a listening test at Swedish Radio, supported by the BBC and some other broadcasters, to choose which were the best audio coding algorithms available. The two "winning" algorithms were called MUSICAM and ASPEC, and these went on to become MP2 and MP3 (official names: MPEG Audio Layer 2 and Layer 3) respecitvely. MP2 was targeted at stereo bit rate levels of 192 - 256 kbps and MP3 was targeted at bit rate levels of 128 kbps -- the trade-off was that MP3 was slightly more computationally complex than MP2.

After apparently "much deliberation" following the listening tests at Swedish Radio, MP2 was chosen as the audio codec to be used on DAB.

Following the selection of the MP2 audio codec, the error correction coding for the system was then designed, and by early 1991 all of the main component technologies -- audio codec, modulation and error correction coding -- had been designed and chosen for DAB, and these main technologies remain unchanged to the present day.

Based on what I've read about the early design of the DAB system in the late 1980s and very early 1990s, my conclusion is that it was rushed and they made mistakes, although they were in uncharted territory, so some (minor) mistakes can be forgiven. The reason I say the design was rushed was because I read that when they chose which modulation scheme to use they had a choice between OFDM and CDMA (CDMA is used by the 3G mobile phone system), and they chose to use OFDM simply because silicon chips were available that could process OFDM, even though simulation results showed that CDMA performed better than OFDM and they would have only had to wait a few months for the CDMA chips to arrive. As it turns out, the choice of OFDM was a good one, but this was a ludicrous way to choose a technology when you consider that broadcasting systems are meant to operate for so many years -- and it becomes even more ludicrous when you consider that DAB wasn't properly launched for over a decade.

The general mistakes they made in terms of design were that they whenever they had a choice they chose to use the least computationally complex technologies: they chose MP2 because it was less complex than MP3; they chose simple error correction coding; and they chose to use differential modulation when they could have used synchronous modulation. The downside of these choices were that the performance of the system was reduced as a result: MP2 requires the use of far higher bit rate levels than MP3; the simplistic error correction coding provides poor reception quality unless the signal strength is high and it leads to multiplexes having a low data capacity; and using differential modulation rather than synchronous modulation means that the transmitter power levels must be twice as high as when synchronous modulation is used.

The reason they chose to use the least complex technologies was mainly because microprocessors weren't very powerful at the time. But all that really shows is that the technology wasn't ready to design a digital radio system -- the fact that it wasn't properly launched for another 10 - 12 years from the time when they were choosing the technologies shows absolutely clearly that there was no inherent consumer demand for digital radio, and when it was finally launched they clearly felt it needed an enormous advertising blitz for it, so they weren't even confident that it would take off when they did launch it.

What they should have done was to adopt MP3 as well as MP2 (because MP3 decoders were originally intended to be backwardly compatible with MP2 -- albeit that most these days cannot decode MP2 -- as they share some of the same processing blocks) and have stronger error correction coding as an option and then let the broadcasters decide whether they wanted to use MP2 or MP3 and stronger or weaker error correction coding. If they couldn't do this with the existing technologies then they should have waited until they could do it.