Digital Audio Broadcasting

DAB uses a wide-bandwidth broadcast technology and typically spectra have been allocated for it in Band III (174–240 MHz) and L band (1.452–1.492 GHz), although the scheme allows for operation between 30 and 300 MHz. The US military has reserved L-Band in the USA only, blocking its use for other purposes in America, and the United States has reached an agreement with Canada to restrict L-Band DAB to terrestrial broadcast to avoid interference.[12]

DAB historically had a number of country-specific transmission modes (I, II, III and IV).

• Mode I for Band III, Earth
• Mode II for L-Band, Earth and satellite
• Mode III for frequencies below 3 GHz, Earth and satellite
• Mode IV for L-Band, Earth and satellite

In January 2017, an updated DAB specification (2.1.1) removed Modes II, III and IV, leaving only Mode I.

From an OSI model protocol stack viewpoint, the technologies used on DAB inhabit the following layers: the audio codec inhabits the presentation layer. Below that is the data link layer, in charge of statistical time division multiplexing and frame synchronization. Finally, the physical layer contains the error-correction coding, OFDM modulation, and dealing with the over-the-air transmission and reception of data. Some aspects of these are described below.

DAB uses the MPEG-1 Audio Layer II audio codec, which is often referred to as MP2 because of the ubiquitous MP3 (MPEG-1 Audio Layer III).

The newer DAB+ standard adopted the HE-AAC version 2 audio codec, commonly known as 'AAC+' or 'aacPlus'. AAC+ uses a modified discrete cosine transform (MDCT) algorithm,[9][10] and is approximately three times more efficient than MP2,[13] which means that broadcasters using DAB+ are able to provide far higher audio quality or far more stations than they could with DAB, or a combination of both higher audio quality and more stations.

One of the most important decisions regarding the design of a digital radio broadcasting system is the choice of which audio codec to use because the efficiency of the audio codec determines how many radio stations can be carried on a fixed capacity multiplex at a given level of audio quality.

Error-correction coding (ECC) is an important technology for a digital communication system because it determines how robust the reception will be for a given signal strength – stronger ECC will provide a more robust reception than a weaker form.

The old version of DAB uses punctured convolutional coding for its ECC. The coding scheme uses unequal error protection (UEP), which means that parts of the audio bit-stream that are more susceptible to errors causing audible disturbances are provided with more protection (i.e. a lower code rate) and vice versa. However, the UEP scheme used on DAB results in there being a grey area in between the user experiencing good reception quality and no reception at all, as opposed to the situation with most other wireless digital communication systems that have a sharp digital cliff, where the signal rapidly becomes unusable if the signal strength drops below a certain threshold. When DAB listeners receive a signal in this intermediate strength area they experience a "burbling" sound which interrupts the playback of the audio.

The DAB+ standard incorporates Reed-Solomon ECC as an "inner layer" of coding that is placed around the byte interleaved audio frame but inside the "outer layer" of convolutional coding used by the original DAB system, although on DAB+ the convolutional coding uses equal error protection (EEP) rather than UEP since each bit is equally important in DAB+. This combination of Reed-Solomon coding as the inner layer of coding, followed by an outer layer of convolutional coding – so-called "concatenated coding" – became a popular ECC scheme in the 1990s, and NASA adopted it for its deep-space missions. One slight difference between the concatenated coding used by the DAB+ system and that used on most other systems is that it uses a rectangular byte interleaver rather than Forney interleaving in order to provide a greater interleaver depth, which increases the distance over which error bursts will be spread out in the bit-stream, which in turn will allow the Reed-Solomon error decoder to correct a higher proportion of errors.

The ECC used on DAB+ is far stronger than is used on DAB, which, with all else being equal (i.e., if the transmission powers remained the same), would translate into people who currently experience reception difficulties on DAB receiving a much more robust signal with DAB+ transmissions. It also has a far steeper "digital cliff", and listening tests have shown that people prefer this when the signal strength is low compared to the shallower digital cliff on DAB.[13]

Immunity to fading and inter-symbol interference (caused by multipath propagation) is achieved without equalization by means of the OFDM and DQPSK modulation techniques. For details, see the OFDM system comparison table.

Using values for Transmission Mode I (TM I), the OFDM modulation consists of 1,536 subcarriers that are transmitted in parallel. The useful part of the OFDM symbol period is 1 millisecond, which results in the OFDM subcarriers each having a bandwidth of 1 kHz due to the inverse relationship between these two parameters, and the overall OFDM channel bandwidth is 1,537 kHz. The OFDM guard interval for TM I is 246 microseconds, which means that the overall OFDM symbol duration is 1.246 milliseconds. The guard interval duration also determines the maximum separation between transmitters that are part of the same single-frequency network (SFN), which is approximately 74 km (45 miles) for TM I.

OFDM allows the use of single-frequency networks (SFN), which means that a network of transmitters can provide coverage to a large area – up to the size of a country – where all transmitters use the same transmission frequency. Transmitters that are part of an SFN need to be very accurately synchronised with other transmitters in the network, which requires the transmitters to use very accurate clocks.

When a receiver receives a signal that has been transmitted from the different transmitters that are part of an SFN, the signals from the different transmitters will typically have different delays, but to OFDM they will appear to simply be different multipaths of the same signal. Reception difficulties can arise, however, when the relative delay of multipaths exceeds the OFDM guard interval duration, and there are frequent reports of reception difficulties due to this issue when there is a lift, such as when there's high pressure, due to signals travelling farther than usual, and thus the signals are likely to arrive with a relative delay that is greater than the OFDM guard interval.

Low power gap-filler transmitters can be added to an SFN as and when desired in order to improve reception quality, although the way SFNs have been implemented in the UK up to now they have tended to consist of higher power transmitters being installed at main transmitter sites in order to keep costs down.

An ensemble has a maximum bit rate that can be carried, but this depends on which error protection level is used. However, all DAB multiplexes can carry a total of 864 "capacity units". The number of capacity units, or CU, that a certain bit-rate level requires depends on the amount of error correction added to the transmission, as described above. In the UK, most services transmit using 'protection level three', which provides an average ECC code rate of approximately ½, equating to a maximum bit rate per multiplex of 1,184 kbit/s.

Various different services are embedded into one ensemble (which is also typically called a multiplex). These services can include:

• Primary services, like main radio stations
• Secondary services, like additional sports commentaries
• Data services
  • Electronic Programme Guide (EPG)
  • Collections of HTML pages and digital images (known as 'broadcast websites')
  • Slideshows, which may be synchronised with audio broadcasts. For example, a police appeal could be broadcast with the e-fit of a suspect or CCTV footage.
  • Video
  • Java platform applications
  • IP tunnelling
  • Other raw data

WorldDAB, the organisation in charge of the DAB standards, announced DAB+, a major upgrade to the DAB standard in 2006, when the HE-AAC v2 audio codec[14] (also known as eAAC+) was adopted. AAC+ uses a modified discrete cosine transform (MDCT) algorithm.[9][10] The new standard, which is called DAB+, has also adopted the MPEG Surround audio format and stronger error correction coding in the form of Reed-Solomon coding. DAB+ has been standardised as European Telecommunications Standards Institute (ETSI) TS 102 563.

As DAB is not forward compatible with DAB+, older DAB receivers cannot receive DAB+ broadcasts. However, DAB receivers that will be able to receive the new DAB+ standard via a firmware upgrade went on sale in July 2007. If a receiver is DAB+ compatible, there will be a sign on the product packaging.

DAB+ broadcasts have launched in several countries like Australia, Czech Republic, Denmark, Germany, Hong Kong (now terminated), Italy, Malta, Norway, Poland, Switzerland, Belgium (October 2017),[15] the United Kingdom and the Netherlands. Malta was the first country to launch DAB+ in Europe. Several other countries are also expected to launch DAB+ broadcasts over the next few years, such as Austria, Hungary, Thailand, Vietnam and Indonesia. South Africa began a DAB+ technical pilot in November 2014 on channel 13F in Band 3. If DAB+ stations launch in established DAB countries, they can transmit alongside existing DAB stations that use the older MPEG-1 Audio Layer II audio format, and most existing DAB stations are expected to continue broadcasting until the vast majority of receivers support DAB+.[16]

Ofcom in the UK published a consultation for a new national multiplex containing a mix of DAB and DAB+ services, with the intention of moving all services to DAB+ in the long term.[17] In February 2016, the new national network Sound Digital launched with three DAB+ stations.[18]

Digital multimedia broadcasting (DMB) and DAB-IP are both suitable for mobile radio and TV because they support MPEG 4 AVC and WMV9 respectively as video codecs. However, a DMB video subchannel can easily be added to any DAB transmission, as it was designed to be carried on a DAB subchannel. DMB broadcasts in Korea carry conventional MPEG 1 Layer II DAB audio services alongside their DMB video services.

as of 2017, DMB is currently broadcast in Norway, South Korea and Thailand.

Some countries have implemented Eureka-147 digital audio broadcasting (DAB). DAB broadcasts a single multiplex that is approximately 1,500 kilohertz wide (≈1,000 kilobits per second). That multiplex is then subdivided into multiple digital streams of between 9 and 12 programs. In contrast, FM HD Radio adds its digital carriers to the traditional 270 kilohertz-wide analog channels, with capability of up to 300 kbit/s per station (pure digital mode). The full bandwidth of the hybrid mode approaches 400 kHz.

The first generation DAB uses the MPEG-1 Audio Layer II (MP2) audio codec, which has less efficient compression than newer codecs. The typical bitrate for DAB stereo programs is only 128 kbit/s or less, and as a result, most radio stations on DAB have a lower sound quality than FM, prompting a number of complaints among the audiophile community.[19] As with DAB+ or T-DMB in Europe, FM HD Radio uses a codec based upon the MPEG-4 HE-AAC standard.

HD Radio is a proprietary system from the company IBiquity. DAB is an open standard deposited at ETSI.

DAB can give substantially higher spectral efficiency, measured in programmes per MHz and per transmitter site, than analogue systems. In many places, this has led to an increase in the number of stations available to listeners, especially outside of the major urban areas. This can be further improved with DAB+ which uses a much more efficient codec, allowing a lower bitrate per channel with little to no loss in quality. If some stations transmit in mono, their bitrate can be reduced compared to stereo broadcasts, further improving the efficiency.

Numerical example: Analog FM requires 0.2 MHz per programme. The frequency reuse factor in most countries is approximately 15 for stereo transmissions (with lesser factors for mono FM networks), meaning (in the case of stereo FM) that only one out of 15 transmitter sites can use the same channel frequency without problems with co-channel interference, i.e. cross-talk. Assuming a total availability of 102 FM channels at a bandwidth of 0.2MHz over the Band II spectrum of 87.5 to 108.0 MHz, an average of 102/15 = 6.8 radio channels are possible on each transmitter site (plus lower-power local transmitters causing less interference). This results in a system spectral efficiency of 1 / 15 / (0.2 MHz) = 0.30 programmes/transmitter/MHz. DAB with 192 kbit/s codec requires 1.536 MHz * 192 kbit/s / 1,136 kbit/s = 0.26 MHz per audio programme. The frequency reuse factor for local programmes and multi-frequency broadcasting networks (MFN) is typically 4 or 5, resulting in 1 / 4 / (0.26 MHz) = 0.96 programmes/transmitter/MHz. This is 3.2 times as efficient as analog FM for local stations. For single frequency network (SFN) transmission, for example of national programmes, the channel re-use factor is 1, resulting in 1/1/0.25 MHz = 3.85 programmes/transmitter/MHz, which is 12.7 times as efficient as FM for national and regional networks.

Note the above capacity improvement may not always be achieved at the L-Band frequencies, since these are more sensitive to obstacles than the FM band frequencies, and may cause shadow fading for hilly terrain and for indoor communication. The number of transmitter sites or the transmission power required for full coverage of a country may be rather high at these frequencies, to avoid the system becoming noise limited rather than limited by co-channel interference.

DLS ("Radiotext") and signal strength, shown on the four line display of this Sony Walkman portable DAB receiver

DAB devices perform band-scans over the entire frequency range, presenting all stations from a single list for the user to select from.

DAB can carry "radiotext" (in DAB terminology, Dynamic Label Segment, or DLS) from the station giving real-time information such as song titles, music type and news or traffic updates, of up to 128 characters in length. This is similar to a feature of FM called RDS, which enables a radiotext of up to 64 characters.

The DAB transmission contains a local time of day and so a device may use this to automatically correct its internal clock when travelling between time zones and when changing to or from Daylight Saving.

DAB is not more bandwidth efficient than analogue measured in programmes per MHz of a specific transmitter (the so-called link spectral efficiency), but it is less susceptible to co-channel interference (cross talk), which makes it possible to reduce the reuse distance, i.e. use the same radio frequency channel more densely. The system spectral efficiency (the average number of radio programmes per MHz and transmitter) is a factor three more efficient than analogue FM for local radio stations. For national and regional radio networks, the efficiency is improved by more than an order of magnitude due to the use of SFNs. In that case, adjacent transmitters use the same frequency.

In certain areas – particularly rural areas – the introduction of DAB gives radio listeners a greater choice of radio stations. For instance, in Southern Norway, radio listeners experienced an increase in available stations from 6 to 21 when DAB was introduced in November 2006.

The DAB standard integrates features to reduce the negative consequences of multipath fading and signal noise, which afflict existing analogue systems.

Also, as DAB transmits digital audio, there is no hiss with a weak signal, which can happen on FM. However, radios in the fringe of a DAB signal can experience a "bubbling mud" sound interrupting the audio or the audio cutting out altogether.

Due to sensitivity to doppler shift in combination with multipath propagation, DAB reception range (but not audio quality) is reduced when travelling speeds of more than 120 to 200 km/h (75 to 125 mph), depending on carrier frequency.[5]

The specialised nature, limited spectrum and higher cost of DAB broadcasting equipment provides barriers to unlicensed ("pirate") stations broadcasting on DAB. In cities such as London with large numbers of unlicensed radio stations broadcasting on FM, this means that some stations can be reliably received via DAB in areas where they are regularly difficult or impossible to receive on FM because of interference from unlicensed radio stations.

Mono talk radio, news and weather channels and other non-music programs need significantly less bandwidth than a typical music radio station, which allows DAB to carry these programmes at lower bit rates, leaving more bandwidth to be used for other programs.

However, this led to the situation where some stations are being broadcast in mono; see music radio stations broadcasting in mono for more details.

DAB transmitters are inevitably more expensive than their FM counterparts. DAB uses higher frequencies than FM and therefore there may be a need to compensate with more transmitters to achieve the same coverage as a single FM transmitter. DAB is commonly transmitted by a different company from the broadcaster who then sells the capacity to a number of radio stations. This shared cost can work out cheaper than operating an individual FM transmitter.

This efficiency originates from the ability a DAB network has in broadcasting more channels per transmitter/network. One network can broadcast 6–10 channels (with MP2 audio codec) or 10–18 channels (with HE AAC codec). Hence, it is thought that the replacement of FM-radios and FM-transmitters with new DAB-radios and DAB-transmitters will not cost any more compared with new FM facilities. It is also argued that the power consumption will be lower for stations transmitted on a single DAB multiplex compared with individual analog transmitters.[24]

Once applied, one operator has claimed that DAB transmission is as low as one-nineteenth of the cost of FM transmission.[25]

The reception quality during the early stage of deployment of DAB was poor even for people who live well within the coverage area. The reason for this is that DAB uses weak error correction coding, so that when there are a lot of errors with the received data not enough of the errors can be corrected and a "bubbling mud" sound occurs. In some cases a complete loss of signal can happen. This situation has been improved upon in the newer DAB+ version that uses stronger error correction coding and as additional transmitters are built.

As with other digital systems, when the signal is weak or suffers severe interference, it will not work at all. DAB reception may also be a problem for receivers when the wanted signal is adjacent to a stronger one. This was a particular issue for early and low cost receivers.

A common complaint by listeners is that broadcasters ‘squeeze in’ more stations per ensemble than recommended [21] by:

• Minimizing the bit-rate, to the lowest level of sound-quality that listeners are willing to tolerate, such as 112 kbit/s for stereo and even 48 kbit/s for mono speech radio (LBC 1152 and the Voice of Russia are examples).
• Having few digital channels broadcasting in stereo.

The nature of a single-frequency network (SFN) is such that the transmitters in a network must broadcast the same signal at the same time. To achieve synchronization, the broadcaster must counter any differences in propagation time incurred by the different methods and distances involved in carrying the signal from the multiplexer to the different transmitters. This is done by applying a delay to the incoming signal at the transmitter based on a timestamp generated at the multiplexer, created taking into account the maximum likely propagation time, with a generous added margin for safety. Delays in the audio encoder and the receiver due to digital processing (e.g. deinterleaving) add to the overall delay perceived by the listener.[5] The signal is delayed, usually by around 1 to 4 seconds and can be considerably longer for DAB+. This has disadvantages:

• DAB radios are out of step with live events, so the experience of listening to live commentaries on events being watched is impaired;
• Listeners using a combination of analogue (AM or FM) and DAB radios (e.g. in different rooms of a house) will hear a mixture when both receivers are within earshot.

Time signals, on the contrary, are not a problem in a well-defined network with a fixed delay. The DAB multiplexer adds the proper offset to the distributed time information. The time information is also independent from the (possibly varying) audio decoding delay in receivers since the time is not embedded inside the audio frames. This means that built in clocks in receivers can be precisely correct.

DAB can provide savings for networks of several stations. The original development of DAB was driven by national network operators with a number of channels to transmit from multiple sites. However, for individual stations such as small community or local stations which traditionally operate their own FM transmitter on their own building the cost of DAB transmission will be much higher than analog. Operating a DAB transmitter for a single station is not an efficient use of spectrum or power. With that said, this can be solved to some degree by combining multiple local stations in one DAB/DAB+ mux, similar to what is done on DVB-T/DVB-T2 with local TV stations.

Although FM coverage still exceeds DAB coverage in most countries implementing any kind of DAB services, a number of countries moving to digital switchover have undergone significant DAB network rollouts; as of December 2019, the following coverages were given by WorldDAB:[26]

In 2006 tests began using the much improved HE-AAC codec for DAB+. Hardly any of the receivers made before 2008 support the newer codec, however, making them partially obsolete once DAB+ broadcasts begin and completely obsolete once all MP2 encoded stations are gone. Most new receivers are both DAB and DAB+ compatible; however, the issue is exacerbated by some manufacturers disabling the DAB+ features on otherwise compatible radios to save on licensing fees when sold in countries without current DAB+ broadcasts.

Portable DAB/DAB+ and FM receiver, circa 2016. This unit requires two "AA" size batteries. (Headphones not shown).

As DAB requires digital signal processing techniques to convert from the received digitally encoded signal to the analogue audio content, the complexity of the electronic circuitry required to do this is higher. This translates into needing more power to effect this conversion than compared to an analogue FM to audio conversion, meaning that portable receiving equipment will have a much shorter battery life, and require higher power (and hence more bulk). This means that they use more energy than analogue Band II VHF receivers. However, thanks to increased integration (radio-on-chip), DAB receiver power usage has been reduced dramatically, making portable receivers far more usable.

Norway was the first country to announce a complete switch-off of national FM radio stations. The switch-off started on 11 January 2017 and ended on 13 December 2017.[27][28] The 2017 switch-off did not affect some local and regional radio stations. They can continue to transmit on FM until 2027.

The timetable for the closure of FM signals in 2017 was as follows:[29]

• 11 January in Nordland (all radio)
• 8 February in Trøndelag and Møre og Romsdal (NRK Radio)
• 21 April in Trøndelag and Møre og Romsdal (P4, Radio Norge and local radio)
• 26 April in Telemark, Buskerud, Hedmark and Oppland (NRK)
• 16 June in Telemark, Buskerud, Hedmark and Oppland (P4, Radio Norge and local radio)
• 21 June in Sogn og Fjordane, Hordaland, Rogaland and Agder (NRK)
• 15 September in Sogn og Fjordane, Hordaland, Rogaland and Agder (P4, Radio Norge and local radio)
• 20 September in Østfold, Vestfold, Oslo and Akershus (NRK)
• 8 December in Østfold, Vestfold, Oslo and Akershus (P4, Radio Norge and local radio)
• 13 December in Troms and Finnmark (all radio)

• The Federal Government of Belgium has decided to turn off FM two years after 50 percent of the radio consumption becomes digital.[30]
• In Denmark the current media agreement of the Danish Parliament stipulates that the FM broadcasts will be stopped as soon as the proportion of digital radio reception reaches 50 percent.[31] Since 2018 the situation has changed when it had not been planned to switch off FM transmitters, at least not until 2023.[32]
• Switzerland has confirmed its plans to switch off analogue radio services starting in 18 months from now. Public broadcaster SRG SSR will go first in summer 2022, followed by private stations by January 2023.[33]
• Sweden in 2015 suspended its plans to switch off.[34]
• Italy, RAS planned the switchover in the province of Trentino between December 2017 and November 2018, but a national switchover isn't planned.[35]

• ETSI specifications available at ETSI Publications Download Area, pda.etsi.org (this will open ETSI document search engine; to find the latest version of the document enter a search string; free registration is required to download PDF)
• Stott, J. H.; The How and Why of COFDM, BBC Research Development