Planning for the future of digital radio

In November last year, the Department for Culture Media & Sport (DCMS) set up a new Digital Radio Working Group (DRWG) 'task force', which is made up of representatives from the DCMS, Ofcom, broadcasters, transmission companies and manufacturers, and they've been tasked with considering the following:

• What conditions would need to be achieved before digital platforms could become the predominant means of delivering radio?
• What are the current barriers to the growth of digital radio?
• What are the possible remedies to those barriers?

They're also looking at how much it would cost to extend digital radio to universal coverage levels, i.e. to around 99% of the population, which means that they're planning towards FM switch-off. Therefore this planning will be crucial to the long-term future of digital radio. Make a wrong decision now, and the UK will suffer the consequences for a long time to come, just like we have done as a result of the incompetent planning for DAB in the late 1990s.

Four sub-groups have been set up within the DRWG to "consider specific issues in greater detail": Technology & Consumer Proposition; Spectrum; Manufacturers; Mechanism for Growth.

DAB planning in the 1990s

Before I look at the current planning, I think it's instructive to look at the mistakes made in the original planning for DAB in the late 1990s to see what lessons can be learnt. The problems with DAB are as follows

1. the audio quality on DAB is low
2. reception quality on DAB is unreliable
3. the range of stations is poor
4. Ofcom estimates that 90 out of the existing 326 analogue stations will never be able to transmit on DAB
5. DAB is very expensive to transmit
6. The rest of the world turned its nose up at DAB due to it using outdated technologies
7. DAB is not being fitted as standard in cars

Problems 1-6 were all completely avoidable, because they would have been solved by the adoption of the AAC audio codec and Reed-Solomon (RS) error correction coding. Problem 7 is the result of very few countries supporting DAB, along with the fact that DAB+ has been designed and other countries wanting to use that instead. Therefore, problem 7 could also have been avoided if they'd have adopted AAC and RS coding for DAB.

The reason why the above mistakes were made was because the people who were doing the "planning" in the late 1990s were people without a technical background, so they didn't understand the strengths and weaknesses of the technologies available at the time. The experts in audio coding at the BBC R&D department had undertaken listening tests in 1996 and 1998, and both of these tests showed that AAC was twice as efficient as MP2, and yet their views must have been ignored by the BBC executives doing the planning.

BBC national DAB multiplex transmission costs

I think it's also instructive to look at how much the transmission costs are predicted to be for the BBC's national DAB multiplex, which are given in the table below for different levels of population coverage.

(There's also £3.6m per year that the BBC spends on transmitting the local BBC stations on DAB.)

The figures in the above table for the current 86% population coverage level and the 90% level that the BBC has said that it has committed to providing are from a report called "The BBC’s Efficient and Effective use of Spectrum" that Deloitte & Touche wrote for the BBC Trust last year: The figures are on page 48 of the PDF file, and regarding DAB providing universal population coverage it said:

However, I placed a Freedom of Information (FOI) request about this issue a couple of months ago, and the figure the BBC provided me with said that it would cost £38m per annum to provide 95% population coverage. So I would assume that where the Deloitte & Touche report said that it would cost "up to £40m per annum" to provide coverage levels "similar to those of FM", I think they will actually be referring to the same information I've obtained under the FOI. However, I would not describe 95% coverage as being "similar to" the 99% population coverage that FM actually provides, because providing broadcast coverage suffers badly from the law of diminishing returns, and the cost of providing universal coverage spirals as coverage becomes close to 100% population coverage. So moving from the £38m per annum figure for 95% population coverage up to 99% coverage would very likely be extremely expensive. This therefore raises the question: how much would it actually cost to provide 99% population coverage that matches FM, because that's what's required if DAB were to ever replace FM?

Mark Friend, the BBC Controller in charge of digital radio, and also the chairman of the DRWG Technology sub-group, said at a Westminster Media Forum last year that DAB had become "prohibitively expensive" to roll out, and that DAB wouldn't achieve universal coverage, so alternative solutions, such as DRM, Wi-Fi and satellite radio (not to be confused with Sky), would need to be used to provide coverage for those who will not be able to receive DAB. However, from reading the presentation slides from the first DRWG stakeholding meeting, the impression I've got is that DAB was the only system they're considering using for the task of providing universal digital radio coverage, so it will be interesting to see how much of licence-fee payers' money the BBC is planning on squandering on its pet DAB project when alternative systems would be far cheaper to roll out.

DRWG planning today

Technology sub-group

I'm going to focus exclusively on the work that the Technology sub-group has been doing. The group is chaired by Mark Friend, the BBC Conroller in charge of digital radio, and I've obtained a list of people that have attended the meetings up to now:

The group has had presentations given by Frontier-Silicon, the market-leading DAB receiver module manufacturer, and from the Society of Motor Manufacturers.
 

The Digital Radio Working Group is clearly incorrectly named: it should be called the DAB Working Group. This isn't about planning for digital radio, this is purely planning for DAB! Where are all the people from the radio industry who think the future of radio is online? Don't they get a say? Anyone suggesting that the Internet should be used for radio amongst that motley bunch would get lynched -- it's about as pro-DAB a group of people as you'd find at the WorldDMB Forum Christmas piss-up! And where are the experts on the Internet and on mobile phone networks? Why haven't they been called to provide evidence?

And look at how few people there are that have a technical background. Have they learnt nothing from the incompetent DAB planning that took place in the 1990s? As you'll see below, the DRWG Technology sub-group has come up with a list of supposed "drawbacks" with using the Internet and 3G to deliver radio, and yet most of the drawbacks are simply incorrect, or they will become incorrect in the near-future, or they're problems that could easily be solved. I mention a lot of forthcoming technologies in the remainder of this article, and virtually none of these technologies received a mention in the DRWG Technology sub-group presentation slides. We're on the verge of seeing some new technologies that will revolutionise both the Internet and mobile communication systems in terms of their ability to carry both live and on-demand broadcast content, and yet none of these technologies have even been considered.

Considering the make-up of the group and the information they've presented so far, I have to say that the DRWG planning has all the hallmarks of the kind of planning that took place in the late 1990s.

The BBC's bias against Internet radio

The BBC is massively biased in favour of DAB and against Internet radio, as the following points show:

• The BBC has broadcast twenty high-impact TV advertising campaigns for DAB, and it has shown zero TV advertising campaigns for its Internet radio and Listen Again streams
• The BBC has been providing absolutely diabolical audio quality on its Internet radio and Listen Again streams despite having the opportunity to use the AAC+ audio codec to improve the quality for over 4 years now
• The BBC was using a bit rate of just 32 kbps for the Internet streams for Radios 1, 2 and 4 until last year, and this was despite Internet bandwidth costs falling in-line with Moore's Law -- i.e. the cost halve every 18 months

And if that doesn't convince you: the BBC has actually commissioned a company to design a DAB radio that has Wi-Fi built into it so that it can connect to the Internet, but despite having Wi-Fi built-in, listeners will not be allowed to listen to Internet radio stations or even the BBC's Listen Again streams -- only DAB can be received!! How unbelievably biased against Internet radio can you get? It even says in the BBC's Royal Charter that the BBC is supposed to encourage people to take-up new technologies:

Fat chance of that happening at the British Bullying Corporation -- you WILL listen the way they tell you, and you WILL be thankful for it!

DRWG Stakeholder Meeting presentation

The DRWG has recently published presentation slides from its first stakeholder meeting, and the most relevant slides for the Technology sub-group are shown below:

Slide 11

Slide 12

It's interesting that they've put Finland and Sweden in the 'DAB' column above, because Finland switched off all of its DAB transmitters over two years ago, and the Swedish government refused to provide funding to Swedish Radio to roll DAB out nationally, which led to most of the DAB transmitters having to be switched off there as well. So they shouldn't be listed in that table at all. The reality is that no country is going to use DAB now that DAB+ has been designed -- using DAB would be idiotic.

Slide 13

As I'll describe below, most of the points made in slide 13 are simply wrong, and they either show that this planning exercise is just a sham and that they're only considering using DAB/DAB+, or that the people doing the planning simply don't understand the technologies they're supposed to be evaluating.

Internet

Not universal coverage

As far as fixed-line Internet coverage is concerned, Ofcom says that broadband is available to 99% of all UK households -- all that is needed for a household to have "Internet coverage" is a phone line connected to a BT local telephone exchange that is broadband-enabled. 99% is the same percentile coverage that FM provides, so the Internet already has "universal coverage".

An EBU Technical Review article called "Indoor reception of DAB -- consequences for planning and implementation", which measured the building penetration losses at office blocks around central London, concluded that a figure of 15 dB would need to be used for building penetration losses for planning purposes. More normally, a figure of around 12 dB is used for building penetration losses, and this requires that the transmissinon powers be 12 dB higher to offset these losses. A 12 dB increase in power equates to the transmission powers having to be 16 times higher in order to provide indoor rather than outdoor mobile coverage.

A large percentage of the costs of providing universal indoor coverage of digital radio are therefore spent on overcoming these building penetration losses, because far fewer transmitters would be needed if only outdoor mobile coverage was required. And yet 99% of all households already have a phone line providing "Internet coverage" that could be used to receive Internet radio, and once digital TV switchover has been completed in 2012, the BBC's radio stations will have universal coverage as well, because they will be on one of the 'public service multiplexes' that are being specifically intended to provide universal coverage.

If indoor coverage of the BBC's radio stations were provided via a combination of the Internet and digital TV, outdoor mobile coverage could be provided extremely cheaply by paying for the BBC's stations to be carried on one of the forthcoming satellite digital radio systems (not to be confused with Sky or Freesat) that will be using either the DVB-SH or ETSI Satellite Digital Radio (E-SDR) standards. Satellite is far and away the cheapest method of providing very large area coverage, which is exactly what is required to provide universal digital radio coverage. The XM and Sirius satellite digital radio systems in the US already use a very similar kind of high-powered satellite digital radio system as DVB-SH and E-SDR will use, and reports have it that the systems in the US work very well.

At a stroke, this would slash the cost of providing universal digital radio coverage by tens of millions of pounds per year, because it would eliminate the need to build hundreds of terrestrial transmitters. Alternatively, if the decision were taken to use one of the terrestrial digital broadcasting systems -- such as DAB+, DVB-H2 or DRM+ -- to provide outdoor mobile coverage, by eliminating the need to provide indoor coverage it would still vastly reduce the transmission costs due to far fewer transmitters being needed due to the far lower field strength that's required.

To provide reception on portable devices around the home, standard wireless routers could be used to transmit Internet radio streams around the home. And digital TV set-top boxes output stereo audio on their SCART sockets, and you can buy SCART to phono adaptor leads cheaply. A lot of digital TV set-top boxes output digital audio via S/PDIF as well. The audio signal from digital TV set-top boxes could be transmitted around the home using one of the following methods:

• Modified wireless router that has phono and S/PDIF digital inputs and transmits the audio signals around the home using Wi-Fi
• Low-power FM transmitter
• Use an emerging home networking standard, such as Wireless USB

The Sky Gnome portable radio is an example of a device that receives audio wirelessly from a Sky set-top box.

On a related note, Ashley Highfield, the BBC's outgoing Director of Future Media & Technology, wrote the following on a BBC Internet blog recently on the subject of the digital divide:

Making use of the Internet to provide indoor digital radio coverage would lead to increased broadband take-up, so this should be investigated further rather than writing it off as having "Not universal coverage". This is supposed to be a planning exercise where one of the tasks is to look at reaching universal coverage, so saying that the Internet doesn't have it shouldn't even have been listed. DAB doesn't have universal coverage either, but I don't see that being listed as a drawback for DAB.

UK capacity limited

The capacity of the UK Internet is a complete non-issue for Internet radio, because the live radio streams will switch over from using unicast as they do now to being distributed via multicast. The total amount of bandwidth required for a multicast radio stream is simply the bit rate that the stream uses, because only a single stream needs to be carried on any given Internet link. The six biggest UK broadband providers account for 91% of the broadband market, and five out of the top six ISPs are either already providing or are planning to provide IPTV services to their customers, and multicast is a key technology to make live IPTV streaming economically viable:

1 - figures from June 2007

The only top-six broadband provider not listed in the above table is Carphone Warehouse/AOL, because I'm not aware of them being interested in providing IPTV services. However, just like with Tiscali, Sky and Orange, once BT's new 21CN has been rolled out nationally by 2010/11 and it is supporting multicast, it would be in the best interest of Carphone Warehouse to convert its network to support multicast because it would save itself a lot of money in terms of bandwidth costs -- converting routers to support multicast simply requires a software update.

On the subject of whether the UK Internet capacity is "limited" or not, I have to say that it's odd that the DRWG Technology sub-group is saying that capacity is limited when the group is being chaired by Mark Friend, who is a BBC Controller, and yet his boss, Ashley Highfield, the BBC's Director of Future Media & Technology, says that there's bags of capacity available on the Internet in the UK. They can't both be right. It looks to me as if the DRWG has simply latched on to the BBC vs ISP row over the iPlayer bandwidth issue, and they've added this as a "drawback" because they're trying to find excuses not to use Internet radio, when in reality it isn't a drawback at all.

Scare stories about Internet capacity being close to breaking point have been doing the rounds ever since the mid 1990s, because Internet traffic has historically grown at a rate of around 100% per year, as the following graph shows:

And Internet traffic is projected to increase by a factor of four over the period of 2007 and 2011, largely as a result of Internet video taking off in a big way due to the launch of services such as the BBC iPlayer and YouTube; video now accounts for the majority of P2P traffic; and IPTV systems are predicted to grow substantially over the next few years as well.

And in an interesting report co-authored by George Gilder, who writes the Gilder Technology Report, called Estimating the Exaflood, it's predicted that Internet traffic in the US alone could hit one zettabyte per year by 2015:

1 exabyte = 10¹⁸ bytes = 1 billion GB

1 zettabyte = 10²¹ bytes = 1 million million GB == 50 million x contents of US Library of Congress (presumably that's a big library...)

However, although the above charts might look like Internet capacity will be limited due to Internet video growing, the Cisco Systems' report and the report co-authored by George Gilder are only actually predicting that Internet traffic will grow at a rate of between 40% - 50% per year, and it's simply the fact that repeating this year after year ends up producing the very large increases in traffic over time. The weakest link in terms of Internet capacity is the speed of Internet routers, because fibre-optic cables can take any level of bandwidth the electronics can throw at them, so it's the electronics used in the Internet routers that are the limiting factor on Internet capacity, and router speeds have increased in-line with Moore's Law since the mid 1980s, as the following figure shows:

As router speeds are increasing in-line with Moore's Law -- doubling in speed every 18 months -- they are outpacing the 40 - 50% annual growth in Internet bandwidth. There's also a 100 gigabit Ethernet (100GbE -- gigabit Ethernet is used to transmit and receive data over the fibre-optic cables that make up Internet links) standard coming out in the next couple of years to supercede the current 10 GbE standard, which will also help to provide greater capacity on Internet links.

Content delivery network (CDN)

There are also architectural changes that could be made to effectively eliminate the iPlayer bandwidth issue in one fell swoop, such as using a content delivery network (CDN) with caches (hard disk-based storage) being installed inside BT's telephone exchanges, or a little upstream from there, which Anthony Rose, the person in charge of the iPlayer at the BBC, mentioned in an article on The Register website.

The typical route that data takes from the BBC's servers to a broadband user is shown in the following diagram:

The reason BT is included in the above diagram is that BT still operates 61% of all UK broadband lines, most of which it operates on behalf of other ISPs, so most data that broadband users receive has to travel over BT's network first. But whether data travels over BT's network or not, the bandwidth costs are incurred on the 'backhaul', which in the case of delivering iPlayer streams to users consists of getting the data from the ISPs' offices to the telephone exchanges, from where it can be sent via ADSL to the user. So by installing a CDN with caches inside the telephone exchanges this eliminates the need for data to travel over the backhaul, and this would therefore effectively eliminate the iPlayer bandwidth costs.

And a CDN with caches inside the exchanges wouldn't have to be limited to only storing iPlayer files, because the iPlayer TV programmes files only consume a small amount of capacity in comparison to the size of today's hard drives, which can currently store up to 1 TB of data. The following table shows what a 1 TB hard drive could store:

A single 1 TB hard drive could store all of the above content as well as having around 415 GB left over for other video, audio and radio applications. The principle of using a cache on microprocessors and on computer hard drives is that memory accesses can be speeded up by putting frequently-used data in faster memory, and a similar principle applies with a CDN, but here the parameter to optimise is bandwidth not latency (latency would actually be improved as well, but that's not the main issue here), so the caches inside the telephone exchanges should store files that account for the most bandwidth. For example, even if a web page is accessed very frequently, it would likely be far better to store an HDTV programme file that is accessed far less frequently, because a 1-hour HDTV programme file at a bit rate of 10 Mbps consumes 4.5 GB, which is about 22,500 times the size of a typical 200 KB web page.

Using a CDN like the one described above could massively reduce the bandwidth costs incurred by ISPs, because it is effectively exchanging the installation of fibre-optic cables and routers to deal with Internet backhaul traffic for using low-cost hard drives and server equipment in the exchanges. Hard drives are cheap and their capacities have been growing faster than Moore's Law for about the last decade, so they could easily deal with the growth in video traffic that's expected to happen over the next few years.

So, is the capacity of the UK Internet limited? No, but growth in Internet video traffic will require investment in new equipment. But investment in new equipment has been part and parcel of being an ISP ever since ISPs began due to Internet traffic growing so quickly over such a long period of time, so this is hardly a new phenomenon, and they will already have budgeted for traffic growth.

High cost to serve

The cost to serve Internet streams is proportional to the total bandwidth required, and I did a back-of-the-envelope style calculation to see how much bandwidth the BBC iPlayer TV streams were consuming, and the figure I came to was about 45 Gbps during peak-times in the evening. I did that calculation a couple of months ago, and the iPlayer is currently growing at a rate of 25% each month, so it must be hitting closer to 70 Gbps by now. Here's a bar chart of the amount of bandwidth the BBC requires to distribute all of its 11 national radio stations via multicast versus the 45 Gbps that I calculated the iPlayer TV streams were consuming:

If you think I've made an error there, I haven't. The figure I used for the eleven BBC multicast radio streams was 128 kbps per station (enough to provide higher audio quality when using AAC than we're ever likely to receive via DAB+), which comes to 1.408 Mbps bandwidth in total, which is 31,960 times less than the 45 Gbps I calculated for the bandwidth the BBC iPlayer was consuming. The bar for the multicast radio streams in the above chart is simply too small to see.

So is 'high cost to serve' a drawback for Internet radio? In a parallel universe where everything that's big is small and everything that's small is big; yeah, it would be costly to serve Internet radio in that universe, but thankfully not in this one. The cost of rolling out the BBC's national DAB multiplex to 99% of the population would be tiny in that universe as well. Perhaps the BBC should roll out DAB coverage there instead?

Reliability low

I listen to Internet radio daily, both via my computer / hi-fi system and using a Wi-Fi Internet radio, and all of the streams I listen to are highly reliable, so I fail to see how they've come to this conclusion.

The issue they're referring to here is Internet streams 'buffering', which used to be a problem a few years ago when I first started listening to broadband Internet radio streams, but it seems to be very rare today, and the only streams it seems to affect are the small-scale webcaster type Internet stations, whereas the BBC's and GCap's Internet streams never seem to suffer from buffering (I haven't tried the streams from the other UK commercial radio groups). I've discussed the issue of buffering with other people, and I've carried out my own brief experiements, and the problem seems to be almost exclusively down to the server being too busy rather than it being down to congestion on the Internet.

Once again, the switch to using multicast distribution would eliminate any problems in this regard, because the server only has to deliver a single stream of each station, so the issue of how busy the server is doesn't even affect multicast in the slightest. Also, the processing power of microprocessors used in servers increases in line with Moore's Law, so smaller webcasters' streams that are using unicast should become more reliable in future as well if they upgrade the hardware they're using.

Buffering on unicast streams isn't an insoluble problem anyway. If the buffer-size used were large enough you could make the probability of a stream buffering so small that to all intents and purposes the issue would be solved. So all it takes is to find a solution where a large buffer-size could be used but which doesn't require the listener to wait a long time for the buffer to fill. One way of doing this would be to have two servers, with one streaming server that delivering live unicast streams as normal, and another server that sends the previous X seconds of audio to the user's buffer when they start listening (taking advantage of the fact that downloading can be done at a bit rate that is far faster than the stream bit rate, hence speeding up the time before audio can be played back). The buffer size would consist of X seconds' worth of audio, so if X was selected to be large enough it would make the probability of the stream buffering very low indeed. The only downside with this is that the stream being listened to would be X seconds delayed relative to what you would hear on FM. But DAB and radio via digital TV are delayed by a few seconds relative to FM anyway, so I don't see why this would be a problem.

P2P live streaming is also an active research area, and the BBC is participating in an EU-funded research project called P2P Next, which aims to deliver a P2P live TV streaming system to people using set-top boxes -- radio streaming is trivially easy in comparison to streaming TV, because the stream bit rates for radio are far lower. P2P streaming could also be used to enhance the reliability of small-scale webcasters' streams, because it would eliminate the load on the server as well as reducing the webcasters' bandwidth consumption, because it's replaced by peers exchanging data with other peers in the P2P network.

Not portable

Wi-Fi Internet radios are obviously portable. I'll deal with mobile Internet reception along with 3G below.

In-car not viable in near term

The following video shows how to receive Internet radio on a car stereo using Microsoft's Sync entertainment system, which is available as an optional extra in the 2008 version of the Ford Fiesta.
 

That's hardly the most elegant of solutions ever devised, but with mobile broadband booming at the moment due to the 3G networks having rolled out HSDPA on their networks, which has allowed them to offer far faster download speeds, higher bandwidth caps and far better value-for-money mobile broadband packages, I think it's only a matter of time before we start seeing in-car entertainment systems having Internet connections, and I would be surprised if in-car Internet connectivity doesn't become a standard feature on new cars over the next 5 - 10 years.

Can't serve broadcast-sized audiences

This wasn't included on the slides shown above, but a few people in the DAB industry have mentioned this as being a drawback for Internet radio recently, so I'll address this here anyway. Multicast allows an unlimited number of listeners, and the bandwith required is the same no matter whether one listener or all 6 billion people in the world were listening to the same radio station. This is not a problem in the slightest.

Wi-Fi

Coverage focused on hot-spots -- unlikely to gain universal coverage

Unlikely? It's absolutely 100% cast-iron guaranteed that it won't gain universal coverage. Wi-Fi is a short-range wireless technology and it is completely unsuited to the task of providing coverage over large areas, which is what's required to provide universal digital radio coverage. There's a rule of thumb for wireless systems that use OFDM (DAB and the 802.11g and 802.11n flavours of Wi-Fi use OFDM, as well as many other wireless standards), which is that the maximum distance between transmitters in a single-frequency network (SFN) is given by the distance that a radio signal travels during the OFDM guard interval being used (presumably they'd want to use SFNs rather than Wi-Fi's slow frequency handovers that wouldn't be able to cope with receivers travelling at high speed). The 802.11g flavour of Wi-Fi uses an OFDM guard interval of 800 nanoseconds, and radio signals travel at the speed of light, which is 3 x 10⁸ m/s, so the rule of thumb dictates that 802.11g transmitters should be separated by 3 x 10⁸ x 800 x 10^-9 = 240 metres at most.

Mobile phone networks use hexagonal cells to simplify frequency planning, so using the same planning methodology as the mobile phone networks, if Wi-Fi were used to provide universal coverage over the 244,820 km² of the UK, I calculate that it would require around 1,634,749 Wi-Fi transmitters. The transmission costs for the BBC's national DAB multiplex are roughly proportional to the number of transmitters, and the current 96 transmitters costs £6m per annum, so, assuming that each Wi-Fi transmitter costs the same amount of money as on DAB (it wouldn't really, but humour me), then in order to provide universal coverage using Wi-Fi it would cost £102.2 billion per year. Just under the size of Jenny Abramsky's pension pot.

However, using Wi-Fi to provide universal digital radio coverage is still an extremely well thought out and sensible plan compared to the original decision to use the DAB system.

Providing universal digital radio coverage

Rather than discussing the drawbacks the DRWG Technology sub-group listed for 3G here, I'm going to write a longer article on the subject of providing universal digital radio coverage, so I'll address them in that, and I'll just provide a relatively brief discussion here of some of the mobile technologies and digital broadcasting systems that should be considered for this task.

Mobile broadband

3G isn't really the right term to use any more, because ever since the 3G networks rolled out HSDPA the new term being used for mobile data transmission is 'mobile broadband'. And just like the case of multicast transforming the proposition of delivering live Internet streams over the wired Internet, there are technologies in the pipeline that will absolutely transform mobile broadband over the next 5 - 10 years, as the following table shows:

Mobile broadband is booming at the moment due to people buying USB dongles to allow them to connect their laptops to the Internet wherever they are rather than being limited to where there's Wi-Fi coverage.

A side-effect of the download speeds increasing enormously over the coming years is that it will enable the 3G network operators to provide much better value-for-money mobile broadband packages in future, because the capacity of their networks will increase massively, and this simultaneously reduces their cost per bit transmission costs.

The key technology that is enabling all of the technologies listed in the table above apart from HSDPA is called MIMO, which stands for 'multiple-input multiple-ouput', and this refers to the fact that more than one antenna is used at both the transmitting and receiving ends of the link. MIMO multiplies the capacity that can be provided relative to using single antennas, and the factor increase in capacity is equal to the minimum of the number of antennas that are used at either end of the link. For example, the 4G prototype system that NTT DoCoMo has designed, and which has been demonstrated transmitting at 5 Gbps to a mobile receiver, is using a 12x12 MIMO antenna array system.

As well as providing higher capacity, MIMO also provides far more robust reception quality than is possible with single antenna systems, because the probability of all the multipaths being faded is vastly reduced due to there being far more multipaths that are being received due to the use of more antennas -- there's a separate multipath channel between each of the antennas at the transmitter and each of the antennas at the receiver, so there are effectively M x N channels, where M is the number of transmitting antennas and N is the number of receiving antennas.

And on the subject of mobile broadband coverage, T-Mobile and Three have agreed a joint partnership to extend coverage of mobile broadband to 98% of the population recently by using network sharing.

Mobile broadband is also the only system that could carry both live and on-demand streams, so considering that we're moving towards an on-demand world, mobile broadband will definitely play a major part in the future of digital radio no matter what, so it would be sensible to combine live and on-demand rather than using two separate mobile systems. Mobile broadband will also obviously be available on everybody's mobile phones over the coming years due to the frequency at which people replace their handsets.

Listing the drawbacks with 3G that the DRWG Technology sub-group identified:

• High cost to serve
• Not universal coverage
• Reliability low
• High cost to end user (at present)

The new mobile broadband technologies we will see introduced over the next 5 - 10 years will solve all of those drawbacks, and I simple don't think they've looked into future mobile technologies anything like enough -- the same can be said for them ignoring the existence of multicast, which the BBC has been testing for serveral years! We're on the verge of a mobile communication systems revolution here, and it would be outrageous to simply ignore that this revolution is about to take place.

There are issues with high battery consumption when streaming to mobiles at the moment, but this could be solved in future by the mobile networks identifying that the data is a stream and then using a lower error correction code rate and/or more robust modulation type (HSDPA uses adaptive coding and modulation, which varies the parameters just mentioned), which would save the mobile handset having to continually update the base station about what coding and modulation to use, which will be the main reason why battery consumption is relatively high -- using the W-CDMA 3G system, mobile handsets have to update the base station 1500 times per second to instruct the base station whether to increase or decrease the transmission power, and mobile handsets transmitting data consumes far more power than merely receiving data. An alternative solution to this problem would be to use time-slicing, which is a technology used by DVB-H to reduce receiver power consumption, where a large chunk of data is transmitted at high speed and the receiver then goes to sleep until it's time to receive the next chunk of data.  This could even be combined with automatic repeat requests (ARQ) for data that's received with too many errors. Basically, this is not a problem that couldn't be solved. And the same applies to reception quality issues as well.

The 3G standard also has a broadcast mode that goes with it, called MBMS (Multimedia Broadcast Multicast Service), so I think it's inevitable that once the 3.9G / 4G standards come out there will also be a broadcast standard to go with these new systems. Considering that the NTT DoCoMo 4G prototype system is 250-times more efficient than DAB and about 80-times more efficient than DAB+, it's not difficult to see that any broadcast standard that's attached to the 3.9G / 4G systems would also be vastly more efficient and therefore far cheaper to provide universal coverage with than DAB+ would be.

Planning for the future of digital radio without properly investigating these future mobile technologies that will transform mobile communication systems over the coming years would be an absolute disgrace.

DVB-SH & E-SDR satellite digital radio systems

As mentioned in the section on the Internet providing universal coverage above, these satellite digital radio systems that use high-power satellite signals would be the cheapest way for the broadcasters to provide very large area outdoor mobile coverage to complement the Internet and digital TV providing indoor coverage, because they would merely have to pay the satellite system provider for carriage.

DVB-H2

DVB-H2 is the forthcoming standard that will supercede the DVB-H mobile TV standard, and just as DVB-H used the same transmission scheme as DVB-T but with some extensions to enable reception of mobile TV, DVB-H2 is expected to use the same technologies and transmission parameters that will be available on DVB-T2 (DVB-T2 is the successor to DVB-T, and it will be used for the Freeview multiplex that will carry HDTV channels), but again with some extensions for mobile TV reception. Assuming that all of the options available to DVB-T2 will also be available for DVB-H2 (the DVB-H2 standard is currently under development), the following list are the key features that are relevant to digital radio:

• MIMO or MISO antenna array technologies (MIMO is not available in DVB-T2, but it is definitely expected to be used in DVB-H2)
• LDPC (low-density parity check) + BCH error correction coding
• 1.7 MHz bandwidth channels -- the minimum channel bandwidth for DVB-H was 5 MHz

As mentioned above, MIMO allows the capacity to be increased whilst providing far more robust reception quality, and LDPC coding is one of the near-optimal forms of error correction coding that have emerged in the last 15 years. The use of both of these technologies would translate into DVB-H2 requiring far lower SNR levels than DAB+ would require, so its use would reduce the cost of a transmitter network due to fewer transmitters being required. When using MIMO it would be possible to use fractal antennas, which are miniaturised antennas that have dimensions of about 50 x 10 x 5 mm.

The addition of 1.7 MHz channels for DVB-H2 also means that it could be transmitted in DAB Band III channels, whereas the narrowest channel bandwidth specified for DVB-H was 5 MHz, which made DVB-H less suitable for local digital radio multiplexes due to spectrum usage issues, and it also increased the transmission powers required by a factor of 3.

DRM/DRM+

Although DRM was listed as a possible complementary technology by the DRWG Technology sub-group, DRM radio stations have been using bit rate levels of around 20 kbps in 9 kHz channel bandwidth up to now, and the audio quality is even worse than it is on DAB, which is saying something. DRM therefore should not be given any consideration due to the exceptionally low audio quality it provides.

DRM+ doesn't have the same problem, though, because a 50 kHz bandwidth DRM+ multiplex could carry a 96 kbps AAC radio station, which would provide good audio quality. DRM+ has a number of advantages over using DAB+ for providing universal digital radio coverage:

• DRM+ is specified to use transmission frequencies between 30 - 120 MHz, and lower frequencies have much better radio propagation characteristics, so far fewer transmitters would be needed to provide universal coverage if DRM+ were used rather than DAB+
• DRM+'s use of lower transmission frequencies also reduces the transmission powers required due to Friis' free space path loss equation showing that the transmission frequency affects the path loss by a factor of (f₂ / f₁)². For example, if DRM+ used a transmission frequency of 100 MHz instead of 200 MHz, which is approximately the frequency DAB/DAB+ uses, the path loss would be reduced by a factor of 4, so the transmission powers would also be reduced by a factor of 4, which would reduce the transmitter network costs.
• DRM+ is far more spectrally efficient than DAB+, because a 50 kHz channel can carry a 96 kbps radio station, so the spectral efficiency is 1.92 bits/s/Hz. DAB+, on the other hand, has a spectral efficiency of just 0.635 bits/s/Hz, which is 3-times lower than the spectral efficiency for DRM+. This also reduces the transmission powers by a factor of 3, because with OFDM-based systems the transmission power is proportional to the channel bandwidth, and it's therefore inversely proportional to the spectral efficiency.
• The combination of DRM+'s use of lower frequencies and higher spectral efficiency reduces the transmission powers relative to DAB+ by a factor of 12, with all else being equal (this doesn't include the superior radio propagation characteristics at lower frequencies)
• DRM+ is geared towards carrying fewer stations per multiplex than DAB+ and other digital broadcasting systems, which makes it more suitable than the other systems for providing a smaller number of stations in an area, and also for providing small coverage areas.
• By combining multiple DRM+ multipliexes together side-by-side, they could be transmitted using single transmitters rather than requiring separate transmission. This would be simple to do as DRM+ channels are only 50 kHz wide, and it would be trivial for a modern-day DSP chip to carry out the modulation for multiple channels.

The DRM+ specification should be released this year, and receivers are due to be available next year. DRM+ therefore has the advantage that receivers could be released at an earlier time than for DVB-H2, which is still under development, but the DVB-SH and E-SDR systems are already specified, and DVB-SH receiver modules are already starting to appear.

Conclusion

The DRWG Technology sub-group seems to have specifically looked for excuses not to use other technologies, and I feel that the people involved in the planning of digital radio, especially the BBC people, have an agenda, and that agenda is that they want to provide universal coverage using DAB/DAB+.

Considering the huge costs involved in providing universal coverage, it would be an absolute outrage if the planning for the future of digital radio didn't thoroughly investigate all of the technologies that could be used, especially as we're at a point in time where we're on the verge of seeing some revolutionary technologies being introduced over the next few years. Ultimately, if DAB+ were used to provide universal digital radio coverage, it will be made obsolete within a short period of time by 4G mobile broadband and 4G mobile broadcast technologies. Is it sensible to use DAB+? I don't think it is, and I don't think the DRWG Technology sub-group have looked anything like far enough into the future to consider what the mobile and fixed-line technology landscape will be like in 10 years' time, because their presentation slides seem to have concentrated solely on the here and now.

It will be many, many years before FM could be switched off, because it's going to be a far more challenging proposition than switching off analogue TV due to there being high demand for more channels on TV, whereas that doesn't seem to be anything like the case with digital radio, because the vast majority of people seem to be happy with what they already receive on FM. Therefore, rushing into decisions that could cost £1 billion or more of licence-fee payers' money over the duration that such a system would operate is simply completely unacceptable.

Bookmark with:
Slashdot	Digg	del.icio.us	Reddit