User:Ryan Cooley/DAB

From Citizendium, the Citizens' Compendium
Jump to: navigation, search

Rough Draft. Probably take a month to look like anything.

Digital Audio Broadcasting (DAB, or EUREKA-147) is a standard for transmitting lossy encoded digital audio.


The EUREKA (European Research Coordination Action) project is a pan-European, intergovernmental initiative, for industrial (non-military) research and development. It does not have any association with the European Union (EU), even though the two share many common goals. [1] [2]

DAB has been in development since 1981, initially by Institut für Rundfunktechnik (IRT).

In 1987 the E!147 project started, taking 6 years to define the DAB standard, and several more to implement it. The DAB standard was publish February 1995, and experimental broadcasts began a few months later (such as the BBC in London). The project ended in 2000, costing a total of 89.2 million Euros. Germany and France contributing the majority, at 36% each. [3]

DAB, DAB+ and DMB are all part of the Eureka-147 family of standards.

"The EBU and the Eureka 147 DAB (Digital Audio Broadcasting) project set up a joint group in 1992, in order to evaluate the performance of the Eureka 147 DAB system."

So the WorldDAB Forum decided in June 2005 to start the development of an alternative audio system for DAB ­ the Technical Committee set up the Task Force, New Audio System. The result of 1.5 years of enthusiastic work ­ the norm "Transport of Advanced Audio Coding (AAC) audio" was published by ETSI in February 2007 and was announced publicly as DAB+ at the same time.


DAB+ was published in February 2007




Canada (may go to HD Radio) UK


Australia Malta Switzerland


South Korea



 Multipath, Doppler Shift, Interference.
 Inter-Symbol Interference (ISI)
 Fast Fourier Transform (FFT)
 Differential QPSK 
   4 quad. phases (0°, 90°, 180°, 270°.)
   no channel equalization needed
   (up to) 1536 (narrowband) carriers
 Guard Interval 
   1/4th symbol length interval
     copies end of symbol before start of symbol
   Single-Frequency Networks (FSN)
   Null symbol
   Phase Reference Symbol (PRS)

Frequency-Domain Multipath

 (Punctured) Convolutional Coding (COFDM)
   Forward-error correction (FEC)
   Normally; half-rate. == 1.2Mb/s
   Viterbi decoder (put the digital signal in the correct chronological order and check the signal for transmission errors)
 Bit-stream re-ordering; Pre-determined patterns
   Temporal:  Between Frames; depth 360ms
   Frequency: Interleaving.  

The performance of this modulation scheme in various channels is described in ETSI 101 758: Digital Audio Broadcasting (DAB); Signal strengths and receiver parameters; Targets for typical operation. For an MPEG-1 Layer II codec, the C/N for non-audio impairment is 14 dB for the majority of locations in urban and rural environments and at speeds below 130 km/h.[4]

-15 db average loss indoor reception[4]

UEP 3 (~0.5)

4 transmission modes, 1,2,4,8KHz wide

 Mode 1: Band III (4x1.54MHz DAB in 7MHz UKTV space)  
   70km max SFN antenna spacing
 Mode 2: L-Band (1452-1492MHz, world) 
   est 17km max SFN antenna spacing
 Mode 3: up to 3GHz for satellite
   est 8km max SFN antenna spacing
 Mode 4: L-Band, sometimes appropriate (L-Band SFNs)
   est 35km max SFN antenna spacing


 The gross data capacity for the entire DAB signal is approximately 3 Mbit/s, of which the Main Service Channel occupies approximately 2.3 Mbit/s. After allowing for the redundancy provided by the channel encoding, a net useful payload in the range of 0.6 - 1.7 Mbit/s is available. For the BBC's national multiplex, the useful payload of the MSC is approximately 1.2 Mbit/s. [6]

6 x 192Mb/s MP2 = 1.2Mb/s Fast Information Channel (FIC) - Non-interleaved

 Multiplex configuration information 
 reference frequency and timing information
 overhead and control. 
 Multiplex Configuration Information (MCI).

?Service Information (SI) Main Service Channel (MSC)

 MP2 audio

Packet Demux for datacasts 24ms frames

Data Services

"ETSI specification, EN 300 401 (Second Edition) specifies the transmitted DAB signal."

'CD quality' using bit-rates of 192 kbit/s or above for stereo

Transmission mode I is intended for terrestrial Single Frequency Networks (SFN) and loca-area broadcasting in Bands I, II and III. Transmission mode II is intended to be used for terrestrial local-area broadcasting in Bands I, II, III, IV, V and in the 1,452-1,492 MHz frequency band (i.e. L-Band).

The DAB signal comprises a succession of transmission frames of 96 ms duration in mode I and of 24 ms duration in mode II. Within these transmission frames, the synchronization channel occupies approximately the first 2.544 ms in mode I, and approximately 0.636 ms in mode II.

(Mode I) ... permits the use of relatively widely-spaced transmitters, whereas in transmission mode II the However, mode II has fewer radiated carriers; also, for mobile reception in L-Band, the greater frequency seperation of the carriers is intended to reduce the effect of Doppler shifting for reception in moving vehicles, especially at the higher speeds. There are 1,536 radiated carriers in mode I, and 384 carriers in mode II, in a system bandwidth of about 1.54 MHz.[7]

Ensemble - A 1.536MHz block of carriers. ETSI - European Telecom Standardization Institute.

Forward Error Correction Levels: Error Protection Level FEC Rate Capacity required for a 192 kbits/s MP2 channel 1 0.34 568 kbits/s 2 0.43 448 kbits/s 3 0.51 384 kbits/s 4 0.62 312 kbits/s 5 0.75 256 kbits/s [8]

Protection level one offers the best protection with a coding rate of approximately 1/3. Level five offers the least protection and has a coding rate of approximately 3/4.

Examples of SI are

  • Service identifier.
  • Programme type.
  • Radio frequencies of associated FM, MF and digital radio services.
  • Announcement switching (linked to announcement channels in an ensemble).

FIC 96 kbit/s (fixed 64 kbits/s ECC) MSC 2.304 Mbit/s

The time and frequency interleaving procedure ensures that bits which are adjacent in time in the sub-channel bit stream are not adjacent in time and frequency when coded onto the 1536 carrier of digital radio signal.

Viterbi or soft-decision maximum likelihood decoding
  Although the states cannot, by definition, be directly observed, the most likely sequence of sets for a given sequence of observed outputs can be computed in O(nt), where n is the number of states and t is the length of the sequence. One method is the Viterbi algorithm.[9]
 By using FFT techniques it is possible to use signals with a special mathematical property: orthogonality. This allows us to pack the separate sub-carriers closely together. Even though their spectra are overlapping it is possible to retrieve the transmitted information without interference.  (DAB The first UK field trial -- BBC 1991-02)
 "Depending on the network scenario and the frequency band used, DAB is three to ten times more spectrum-efficient than FM."[10]


DAB can be said to have a more gradual "digital cliff" effect than most digital broadcast standards. Where reception is weak, and numerous errors in the channel make perfect reception impossible, audible distortion can be heard, rather than the signal being completely lost. The artifacts are said to sound like "bubbling mud".

In many areas, the spectrum being used for DAB is in-use for other (analog) broadcasts, such as TV. This has required DAB broadcasters to use much lower power levels than their analog FM counterparts to avoid interference with existing broadcasts (that haven't yet been obsoleted and "switched-off"), making digital reception relatively more difficult.[11]

DAB is broadcast at a substantially higher frequency than analog FM. Around 100 MHz for FM, versus 1,500 MHz for DAB. Propagation is considerably different at those significantly higher frequencies, so it's inherent than in some situations where analog radio was easily received, DAB cannot be. These higher frequencies are particularly disadvantageous when the signal must travel through objects (like buildings, or even tree leaves) to be received. So DAB is much more sensitive to having "line-of-sight" than is analog radio, through no inherent fault of its own.

 the cost of transmission ... is almost doubled due to simulcasting;
 It is pointless to reserve spectrum for future satellite DAB services since there is little or no commercial interest in pan-European DAB services over the foreseeable future. It should be pointed out that, for large-area coverage using a Single Frequency Network (SFN), VHF is more appropriate than L-Band. The latter seems to be technically more suitable for small-coverage zones, e.g. for small local and community broadcasters.[10]
(DAB) can in principle carry only one stereophonic service. It would be sensible, where geographically possible, to use a dedicated ensemble for a stereo service, with the highest possible audio bit-rate (i.e. 384 kbit/s) and protection level (1/3). This configuration would allow high audio quality and low transmitter power, thus minimizing the installation costs. It would also allow us to shape the coverage of the multiplex, as required by the market. 

Data Coding

MPEG-1 Layer II audio at 48kHz or 24kHz

A 192kbit/s DAB channel can cost from 2 to 20 million Euros, depending on how well utilized each transmitter is.[12]

insert PAD into the Musicam frames

 Electronic Program Guide (EPG)


 "The BBC covers 85% of the UK population" [13]
 "In the United Kingdom, 12.5 MHz of Band III spectrum from 217.5 - 230 MHz has been allocated to DAB. This will accommodate seven multiplexes. The BBC has been allocated one of these channels for its national DAB multiplex" [6]

Commercial DAB licenses "starting in the Spring of 1998". BBC full SFN completion at the same time.[6]

 "RDS travel bulletins to interrupt listening,"

"The application fee for the national multiplex was £50,000 and the Radio Authority is charging an annual licence fee of £10,000."[10]


 "100 million people" circa 2007


 Reed-Solomon additional ECC 8.3% overhead
 Marginally better fringe reception due to R-S
 Muting instead of artifacts
 Stream and Packet Mode. The ten parity bytes per 110 data bytes ­ equivalent to an over-head of 8.3%­ lead to an ability of correcting up to five erroneous bytes in those 120 bytes (Fig. 9).

 DAB+ was published in February 2007 as ETSI TS 102563 "Digital Audio Broadcasting (DAB); Transport of Advanced Audio Coding (AAC) audio".
 HE-AAC v2 provides the same perceived audio quality at about one third of the sub-channel bitrate needed by MPEG Audio Layer II.  DAB+ Bullshit[14]

Moser AAC+ Bullshit. Does not mention MP2 at all[15] Ridiculously low AAC bitrates are cited are based on ITU-R BS. 1534 MUSHRA (MUltiple Stimulus test with Hidden Reference and Anchor) testing, versus ITU-R BS.1116-1 testing for older, higher MP2 bitrates.[16]

For traditional DAB, SBR can be combined with MP2 to provide better audio quality at a given bitrate (to newer receivers equipped with SBR decoding), while legacy decoders would still be able to listen to a lower quality audio signal.[17] This would eliminate the need for the 8% overhead of reed-solomon codes, and maintain backwards compatibility with DAB receivers already sold. Tests by IRT have shown that DAB reception is not affected by SBR. [18]

DAB will actually outperform DAB+ in some cases... MPEG-1 Layer II performs just as well as AAC in high quality/high bitrate audio encoding (See: MPEG-1#Quality) yet doesn't require the additional 8.3% overhead of Reed-Solomon error-correcting codes DAB+ added to make AAC usable. DAB+ also has much higher computational requirements.

In areas like the UK, where DAB sound quality is being sacrificed by broadcasters to cut costs and include more channels, it's likely DAB+ would be beneficial. In areas where channel demand is not as high, and channel bitrates of 192kbit/s are practical, DAB+ doesn't offer any advantages.


See Also


External Links