The Incompetent Adoption of DAB in the UK

Justification for the use of the word "incompetent"

When you look at the history of radio broadcasting, one of the most striking things is how long the systems have lasted:

• AM radio was first commercially broadcast in 1920
• FM was invented in 1935, there was an FM broadcast band in the US in the 1940s, and the Zenith-GE pilot tone system was standardised in 1961 to provide FM stereo, and FM stereo has remained unchanged up to the present day

DAB on the other hand was "properly" launched in the UK in 2002, yet just 3 years later the WorldDAB Forum pulled the plug on the old DAB system by ordering that the AAC+ audio codec be adopted, which led to the design of the new DAB+ system, which will make all DAB receivers obsolete in the coming years.

Because 3 years is such an extremely short duration in broadcasting system terms, the launch of the old DAB system in the UK has got to go down as the most incompetent technical decision ever made in the history of broadcasting -- and that includes both TV and radio. 

"Incompetent" is a strong word to use, but I'm afraid that in this instance I feel it is perfectly jusified. 

And if you're now thinking "it's easy to say this in hindsight", the technologies had existed for years before DAB was launched in the UK, but the BBC didn't upgrade DAB prior to launching it. For example, the AAC audio codec had been standardised in 1997, and Reed-Solomon error correction coding is used as the error correction on CDs, so it had been in widespread use since the 1980s. If these two technologies had been used to upgrade the DAB system prior to it being launched, the audio quality and the reception quality would be far better than they are with the current system, and DAB would actually be able to carry all of the analogue stations, whereas Ofcom has admitted that around 90 analogue stations will never be able to fit on DAB due to either being unable to afford the sky-high transmission costs or due to the local DAB multiplexes being full.

Of course there have been other broadcasting system failures, but nothing comes close to matching the significance of the launch of a system that was meant to be the digital replacement for the ubiquitous FM system followed just 3 years later by its ruling body scrapping it.

It also should not be forgotten that when the UK launched DAB in 2002 they did so thinking they were pioneering DAB and that the rest of Europe and then most of the rest of the world would naturally follow their lead. But DAB+ was only actually designed because this plan went so incredibly wrong that the only countries the UK could get to commit to using the old DAB system were Denmark and then more recently Norway, with virtually every other country that said anything on the subject of digital radio being opposed to using the old DAB system. 

Put simply, if DAB+ hadn't been designed, the UK, Denmark and Norway would have been the only countries stuck using the old DAB system, while all other countries would have adopted one of the modern and far more efficient systems that can be used to carry digital radio, such as DVB-H, T-DMB, HD Radio or DRM+. And the saddest thing about this whole story is that it was so easily avoidable, for the reasons I will expand upon below.

The "true" launch of DAB in 2002

Although the BBC began transmitting DAB in 1995, there were no DAB receivers on sale at all until Arcam brought out its Alpha 10 tuner in December 1999, which cost £800.

But DAB was only "properly" launched in March 2002 when the BBC launched 6 Music, which was the first of five new BBC digital-only stations to launch that year. But the crucial element that made 2002 the true launch-date of DAB was the beginning of the advertising blitz that coincided with the launch of the BBC's digital-only stations, and as of today (18/4/07) the BBC has broadcast 19 high-impact TV advertising campaigns for DAB, which I've calculated would have cost £155 million¹ if the BBC had to pay for these adverts to be broadcast on commercial TV -- with DAB receiver sales standing at around the 4 million mark, that means the BBC has pseudo-subsidised DAB to the tune of £155m / 4m = £38.75 per DAB radio!

Then just three and a half years later in October 2005, the WorldDAB Forum (now called the WorldDMB Forum) ordered the Technical Committee to do the work necessary to add the AAC+ audio codec, which was a decision that will make all of the existing DAB receivers obsolete in the coming years.

The design of the old DAB system

This section is a short summary of how the old DAB system was designed, but for a slightly more in-depth description see here.

The old DAB system was designed in the late 1980s and early 1990s, and all of the main component technologies -- the audio codec, the modulation and error correction coding -- had been chosen by early 1991, and they remain unchanged to the present day.

The designers of the old DAB system chose to use low complexity technologies, which was mainly due to the fact that microprocessors weren't very powerful in the late 1980s. Examples of this were that the designers chose to use the MP2 audio codec instead of MP3, and they chose to use simple rather than the more complex but stronger error correction coding they could have used.

The downside of choosing to use these low complexity technologies was that they made DAB an incredibly inefficient system. For example, whereas MP3 was designed to be used at 128 kbps, MP2 was designed to be used at bit rate levels between 192 - 256 kbps, and this is borne out by the following quote in a BBC R&D report about DAB from 1994:

And combining the inefficiency of the MP2 audio codec with the weak error correction coding used on DAB (weak error correction coding leads to DAB multiplexes having a low data capacity -- i.e. a low spectral efficiency), DAB multiplexes can only carry a very small number of radio stations:

1 - the bandwidth of a DAB multiplex is 1,710 kHz

DAB at the BBC -- 1990 - 2002

The following bullet points are a time-line of some relevant events to do with DAB at the BBC in the 1990s and early 2000s:

• January 1990 - BBC R&D department first trialed DAB by transmitting it from Crystal Palace
• 1991 - BBC R&D demonstrated DAB to press
• 1992 - "Extending Choice: The BBC's Role in the New Broadcasting Age" document was published
• September 1995 - BBC national DAB multiplex began broadcasting Radios 1-5 nationally in 1995 (BBC World Service was added at a later date)
• May 1996 - "Extending Choice in the Digital Age" document published by BBC Director-General John Birt 
• September 1998 - bbc.co.uk mentions BBC's plans to launch 4 digital-only radio stations (BBC Paliament, Asian Network, new music station, new sports station)
• December 1999 - The first DAB tuner went on sale -- the Arcam Alpha 10 tuner -- costing £800
• Summer 2001 - BBC deliberately withholds information that the launch of new stations will drastically degrade the audio quality of existing stations in its public consultation for new digital radio stations (this was admitted in an email in 2002 by the then Controller of Radio & Music Interactive, Simon Nelson)
• November 2001 - Mediocre feedback from consultation (2 of the proposed stations get less than 50% acceptance figures and the only station to get more than 60% acceptance was BBC7)
• 2002 - 5 new digital-only stations are launched (6 Music, 1Xtra, BBC7, Asian Network, Radio 5 Sports Extra)

Soon after the BBC's national DAB multiplex was launched in 1995, it was carrying the following stations:

The BBC will have known from the early 1990s -- straight after the listening tests at Swedish Radio in 1990 that led to the MP2 audio codec being adopted for DAB -- that DAB would need to use a bit rate level of 256 kbps to provide near CD-quality (the provision of near CD-quality was the main reason why DAB was designed in the first place), and yet when the BBC launched its national DAB multiplex in 1995 it obviously couldn't use a bit rate level of 256 kbps or else Radio 5 and the World Service wouldn't have been able to be carried on the multiplex. So, instead, the BBC reduced the bit rate levels of Radios 1-4 to 192 kbps -- i.e. when they launched their national DAB multiplex they had already reduced the audio quality level to below the 224 kbps require to match FM! However, this didn't really matter at the time too much, because you couldn't buy a DAB receiver until December 1999, and then they cost £800.

It is just staggeringly incompetent that the BBC knew from the early 1990s that they would be adding a number of new digital-only stations to their national DAB multiplex, yet they only had enough space left for one additional stereo station, and then in 2002 they actually added 5 new digital-only stations. How long did they need to figure out that DAB wasn't up to the job??

The bit rate levels of all of the BBC's music stations apart from Radio 3 are now half the bit rate levels that were suggested to be used originally and the audio quality is awful.

Number of radio stations available

The majority of people in the UK can currently receive 4 DAB multiplexes, which consists of the BBC and Digital One national multiplexes, a regional and a local multiplex. Once the forthcoming DAB expansion has finished, the majority of people will be able to receive 5 DAB multiplexes due to the addition of a second commercial national multiplex that will launch in 2008. This means that if the recommended bit rate levels of 256 or 224 kbps had been used to provide the near CD-quality or at least FM-quality that DAB was originally designed to provide, people would be able to receive the following number of stations:

1 - the table includes the effect of mono stations using half the bit rate of stereo stations -- 80% of radio stations on DAB are stereo stations and 20% are mono, and this works out as adding one extra station per multiplex whether 256 kbps or 224 kbps is used

The new spectrum for the expansion of DAB is all there's going to be DAB, because this new spectrum was acquired from the Regional Radio Conference (RRC-06) in Geneva last year, and the last time there was a frequency planning conference on that scale was in Stockholm in 1961!

Twenty-five to thirty radio stations -- or 36 stations tops in London -- was never going to be enough, so the above table basically shows that it was inevitable that poor audio quality would be provided on DAB. I would therefore suggest that the BBC and the Radio Authority (the regulators of commercial radio pre-Ofcom) were grossly incompetent to use a digital radio system where it was inevitable that poor audio quality would be provided.

Transmission costs per radio station

I was provided with some actual DAB and FM transmission cost figures once by someone in the DAB industry, which are contained in the following table along with how much it would cost to transmit at 224 kbps or 256 kbps based on the fact that DAB transmission costs are pro rata with the number of capacity units (CU) consumed (capacity units are usually but not always linearly proportional to the bit rate levels):

According to Ofcom, 50% of all existing analogue radio stations make no profit. I would therefore suggest that the above table shows that using DAB makes providing poor audio quality inevitable, because commercial radio stations wouldn't be able to afford to pay the transmission costs to provide good audio quality. The Radio Authority was therefore incompetent to propose that DAB should be used as the digital radio system in the UK.

The above table also goes some way to explain why only around 45% of all existing analogue radio stations are on DAB. The big stations owned by the big commercial radio groups are on DAB -- the big commercial radio groups own the commercial DAB multiplexes -- but the small and medium-sized stations are not on DAB, because most simply cannot afford the transmission costs. DAB is a great way for the big commercial radio groups to monopolise digital radio by excluding access to their competition. 

Ofcom has said that even after the expansion of DAB, 90 out of the existing 326 commercial radio stations won't be able to transmit on DAB either due to not being able to afford it (which probably accounts for most of them) or because the multiplexes will be full.

Better technologies were ready and waiting to be used

When I've criticised DAB in the past numerous people have said "oh, it's easy to see mistakes with the aid of 20/20 hindsight" or words to that effect. The reality is that technologies were sitting there ready and waiting to be used, but those in charge of DAB stuck their heads in the sand, and the people at the BBC, the Radio Authority and in commercial radio were probably too oblivious to there even being a problem.

As I mentioned earlier, the problem with DAB is that it is an extremely inefficient system, so below I'll say which technologies existed that could have vastly increased the efficiency, which would have avoided the current problems altogether.

Reed-Solomon error correction coding -- invented in 1960

Reed-Solomon (RS) coding is used as the error correction coding on CDs, and its efficacy for use in conjunction with OFDM modulation had been shown in an EBU Technical Review article (edition 224) by Alard and Lasalle in August 1987, so it was obviously around when DAB was originally being designed, but they obviously either ignored it or deemed that it was too computationally complex. I'm afraid that if they deemed it to be too computationally complex then they were trying to design a digital radio system before digital processing was fast enough to handle it, and they should have waited until Moore's Law caught up.

RS coding could have increased the multiplex data capacity by approximately 40%. RS coding has now been adopted for DAB+.

MP3

MP3 was the joint-winner with MP2 of the listening test at Swedish Radio in 1990 that led to the designers of DAB adopting MP2, so MP3 obviously could have been adopted at any point during the 1990s. 

MP2 was targeted at bit rate levels between 192 - 256 kbps, whereas MP3 was optimised for use at 128 kbps. However, MP3 at 128 kbps would have provided similar audio quality to 192 kbps MP2, so MP3 was 50% more efficient than MP2.

AAC

Development of AAC began in 1993 when it was shown that superior compression performance could be achieved by removing the requirement for codecs to be backwardly compatible with MP2 and MP3. AAC was standardised in 1997. 

If the designers of DAB had spotted that their system was far too inefficient they could have worked to adopt AAC in parallel to the ongoing development work and incorporate AAC soon after it had been standardised in 1997. And considering that the first DAB receiver didn't go on sale until December 1999, AAC obviously could have been adopted for DAB. 

AAC is twice as efficient as MP2 -- ironically, it was in listening tests carried out by BBC R&D on AAC that proved that AAC was twice as efficient as MP2: first for a multi-channel test in 1996, then for a stereo test in 1998. To be fair to the BBC R&D engineers, they did say in subsequent documents how good AAC was, so presumably the BBC management not having a clue about technology that was to blame for the glaring error.

Combination of more efficient audio and error correction coding

The following table shows the combined effect of using the more efficient audio and error correction coding relative to that used on DAB:

SBR - mp3Pro & AAC+

Spectral Band Replication (SBR) is the technology that makes AAC+ more efficient than AAC, and some of you may remember a codec that was around a few years ago called mp3Pro, which, like AAC+, consists of the addition of SBR, but this time to MP3. So if either MP3 or AAC had been adopted then it would have been relatively painless to add SBR in order to make DAB even more efficient.

Conclusion

If either of the above two options had been adopted instead of the technologies that the old DAB system actually uses then there wouldn't have been a problem, because the broadcasters wouldn't have needed to provide low audio quality in order to provide the number of stations they are providing in the UK. 

However, unfortunately the above two options weren't used, and the audio quality on DAB is low, and it will be low for the next few years. The problem with the audio quality will be sorted out once DAB+ has been fully adopted -- there will be no reason to provide low audio quality then, albeit that I wouldn't put it past some of the most tight-first commercial radio groups continuing to provide the same low audio quality as they do now, in particular GCap Media and Emap, where they actually seem to be proud of providing as low audio quality as possible. 

But it will be several years before all of the legacy MP2 services have been switched off, albeit that we will see DAB+ stations launch within the next 3 years or so. But MP2 and AAC+ services will have to go through a period of being transmitted in parallel before the time comes when incompetent Ofcom allows the MP2 services to be switched off, so the quality on DAB is -- amazingly -- going to get worse before it gets better. 

We've already had 5 years of a sub-standard service, so the effect of the incompetent adoption of DAB in the UK will have lasted a long time before the final MP2 service has been switched off.

1 - I carried out a detailed calculation to work out how much the BBC's 19 high-impact TV advertising campaigns would have cost if the BBC had