CAP 7 PAR2

reception. The effect of multipath reception is depicted graphically in Figure 1.9. Both the direct signal from the transmitter and reflected signáis arrive at the antenna of a receiver.
All signáis contain identical information but arrive at different instances of time. Owing to the measures previously described, the DAB receiver is able to cope with these multipath signáis.
Going a step further, it is now irrelevant for the receiver whether the delayed signáis were originated by the same transmitter or come from another transmitter that transmits exactly the same information synchronised in time as shown in the second part of Figure 1.9. This means that DAB allows coverage of any área with a number of transmitters that transmit the identical programme on the same fre¬quency. Such broadcast networks are called single frequency networks.
The SFN capability of DAB transmission networks can be seen as an extra benefit which is an implicit result of the initial requirement during the development of DAB to cope with multipath phenomena typical of mobile radio reception.
Four different transmission modes were developed to cater for a wide range of speed and frequency requirements in DAB systems as shown theoretically in Figure 1.10. Dependent on the transmit frequency and máximum vehicle speed, the appro-priate DAB transmission mode can be chosen.


Figure 7.11 gives the results of a number of computer simulations, laboratory measurements as well as field measurements which ¿Ilústrate the impact of the Doppler spread caused by the vehicle motion on the required C/N at the receiver. The graph in this figure is expressed in terms of margin above the basic C/N required in the case of slow Rayleigh channel fading to allow reception, as a function of the vehicle speed. Note that the Doppler spread results from different Doppler shifts on



the various multipath signáis arriving at the receiver, according to their direction of arrival relative to the direction of vehicle displacement.
The results were all scaled to the utilisation of DAB transmission mode IV at 1.5 GHz. A simple linear interpolation can be applied to make these results useful to other transmission modes and carrier frequencies. It gives a more refined estímate of the Doppler spread performance of the DAB system at lower vehicle speeds. Two curves give an empirical estímate of the service availability performance at 80% and 99%. A detailed explanation of the mobile radio channel and the DAB transmission scheme can be found in Chapter 2.
7.3.3 Advantages of SFNs
In the following section two main advantages of SFNs over MFNs will be briefly introduced: power economy and frequency economy. A more detailed discussion of these issues can be found in section 7.4.
7.3.3.1 Power Economy of SFNs
Owing to the properties of the system, DAB receivers can use all signáis received in an SFN in a constructive manner. This works as long as all signáis arrive within the guard interval. Signáis with longer delays créate self-interference problems in the SFN and must be avoided by careful network plannmg.
Moreover, SFNs also bring the benefit of space diversity due to the fact that transmitters are located at different locations in the network. The diversity effect, resulting from the fact that the probability of simultaneous shadowing in the pres-ence of several signáis is much lower than the probability for shadowing for a single signal, contributes an additional statistical network gain.
DAB networks are therefore very power economical. The advantages offered by SFN operation and the properties of the digital transmission system itself, allow lower transmitter powers compared to FM for the same coverage quality. Power savings can be as high as lOdB. Figure 7.12 shows this effect in a qualitative manner. To cover the same área with one programme on the same frequency, FM needs one high-power transmitter, whereas DAB with an SFN network with several transmit¬ters needs much lower transmitter power in total. Another effect of the DAB SFN compared to FM networks is the much lower spill-over resulting from the steeper fíeld strength slope at the edge of the coverage área as illustrated in Figure 7.12 which reduces unwanted interference in neighbouring networks
7.3.3.2 Frequency Economy of SFNs
The fact that SFNs allow coverage of large áreas occupying only a single frequency in the spectrum results in highly frequency economical planning possibilities for com-plex broadcast landscapes. The SFN technology also allows successive improvement of coverage quality without having to re-plan frequency allocations. Coverage prob¬lems within the network can be solved by simply putting up additional transmitters. The additional planning that must be done is to check the timing constraints of the


Figure 7.12 Power economy of DAB SFNs in comparison to FM
SFN to avoid violation of the timing budget given by the guard interval of the chosen DAB transmission mode. Note that extra care must be taken to avoid an impact on reception in distant parts of the network because of spillover from the additional transmitter during abnormal propagation conditions.
7.4 Particularities of SFNs
7.4.1 Review of COFDM Principies
As stated earlier in the chapter, COFDM was originally intended to provide success¬ful reception in multipath propagation conditions arising from reflected signáis, but it works equally well for reception of múltiple transmitters carrying the same signáis. This gives rise to the possibility of an SFN, in which all transmitters carry the same information at the same time (or nearly the same time). Important factors for successful implementation of SFNs are the accuracy of the frequency and the timing of each transmitter. In addition, the length of the guard interval is important for an SFN implementation, because it influences the allowable range of transmitter spacing (see also section 7.6.4).
7.4.2 Time and Frequency Synchronisation
For an SFN to opérate effectively, the transmitters must deliver the DAB signal to the receiver at the same time, or nearly the same time, and at the same frequency.


Frequency errors between the transmitters cause a loss of orthogonality between the received carriers, and also reduce the receiver's tolerance to the Doppler spread effects experienced in mobile reception.
Timing errors between transmitters erode the guard interval of the composite received signal, and can therefore disrupt the performance of the SFN.
For these reasons, transmitter networks are normally specifíed to keep transmitter frequencies within 1% or so of the carrier spacing, and timing within a few per cent of the guard interval. In order to achieve this, an independen! and ubiquitous time and frequency reference is required. Global Positioning System (GPS) receivers are commonly used for this purpose. These receivers offer time and frequency references, typically 1 pulse per second (pps) and lOMHz signáis, with an accuracy well in excess of that required for DAB SFNs. The 1 pps signal is used to define the transmission time of the data, and the 10 MHz signal is used as a reference for LO synthesisers that determine the final radio frequency.
7.4.3 Signal Reinforcement from Nearby Transmitters, Network Gain
If two or more transmitters serve the same área, their signal strengths are, in general, not strongly correlated. The signal strength of the transmitters varies with location, but because the signáis are not strongly correlated, an área of low signal strength from one transmitter may be "filled" by higher signal strength from another trans¬mitter. In RDS (Radio Data System, used in FM networks to carry additional information), this is exploited by allowing the receiver to retune to alternative frequencies carrying the same programme if reception of a particular frequency is poor; which is called frequency diversity. However, unless two receiver front-ends are used, the receiver has to retune without prior knowledge of whether the alternative frequencies will offer greater signal strength.
In a DAB SFN, retuning is not necessary. For múltiple transmitters on the same frequency, it remains true that an área of low signal strength from one transmitter may be "filled" by higher signal strength from another transmitter. This is a forra of "on-frequency" diversity, in which the receiver does not have to retune, although it may adjust its synchronisation to make best use of the available signáis.
Another way of looking at this phenomenon is to consider the aggregate strength of the composite signal from a number of transmitters. This varies less with location than the signal strength from any of the individual transmitters, or in statistical terms, the variance of the strength of the composite signal is lower.
This effect offers a major advantage. In a multifrequency network, a number of transmitters may provide signal strength to an área without providing adequate coverage. This is less common in SFNs because of the tendency of two or more transmitters to ful each other's coverage deficiencies. This results in the coverage of an SFN being greater than the sum of the coverages of its individual transmitters, and is often known as "network gain".

7.4.4 Effects of Distant Transmitters
Nearby transmitters on the same frequency have a constructive effect, but in a large SFN, the more distant transmitters, whose signáis may arrive outside the guard interval, can act as interferers. Although it is possible for signáis arriving just beyond the guard interval to contribute some useful energy, they will at the same time contribute to the interfering power and this latter effect will in most cases be predominant. This is a complicated issue infiuenced by the design of the receiver, and therefore the effects are somewhat variable.
Transmitter spacing is therefore a factor in network design, but in practice the availability of suitable transmitter sites, topography and population density are also major influences. For Band III SFNs, using Mode I, transmitter spacings are usually somewhat smaller than the distance corresponding to the guard interval (around 75 km). This may result in many transmitters contributing useful energy under favourable circumstances, but in SFNs above a certain size there will always be potential for interference from distant transmitters. Because of this, SFNs have to be planned taking very careful account of the interference caused by transmitters to distant locations, as well as the coverage provided in their immediate surroundings. More information about máximum transmitter spacing as function of the DAB Mode is given in Table 7.5.
7.4.5 Optimised Coverage of SFNs
Although the transmitters in an SFN need to deliver the signáis with very precise timing, it is not always necessary for them to be exactly co-timed, and in some circumstances it can be advantageous to offset the timing of particular transmitters by significant fractions of the guard interval. This is particularly true at the extrem-ities of coverage, or where low-power transmitters are used to fill gaps within coverage provided primarily by high-power transmitters. Transmitter timing is a variable in network design and can be used in combination with transmitter powers and directional radiation patterns to optimise the coverage of the network.
7.4.6 Closure of Coverage Gaps Using Gap Fillers
In conventional MFNs, as used for FM broadcasting, coverage gaps are closed with additional transmitters, which need their individual frequency assigned. This requires careful planning to ensure that incoming and outgoing interference is correctly managed. This is often not economical from the frequency spectrum point of view. SFNs, however, allow relatively simple filling of áreas not well served by the main transmitters, that is gaps in the coverage área, by installing low power on-channel repeaters located inside the coverage área which opérate on the same frequency as the rest of the SFN (see Figure 7.13). Typical under-served áreas include zones shadowed by natural or man-made obstructions such as valleys, tunnels and in cities blockages behind tall buildings.


Figure 7.13 Typícal application of a DAB gap fíller
These additional re-transmitters with a typical output power of the order of a few watts are called gap fíllers or repeaters. A gap fíller is simple to construct and install since it requires relatively small power and can be mounted on a small tower or on the roof of a building. The receiving antenna of the gap-filler should be highly directional with reduced back lobes, while the re-transmitting antenna will generally be tailored to the specific characteristics of the shadowed área. Gap fíllers must be located at points in the network where there is sufficient incoming field strength and where the re-transmitting antenna can be directed towards the as yet uncovered área oftheSFN.
In the Canadian DAB networks more powerful re-transmitters with an output power of up to 100 W are also used. This type of device is called a coverage extender since it is typically located at the fringe of the network and fires beyond the coverage área of a main transmitter. Coverage extenders enlarge the total área of the SFN instead of serving regions not well covered owing to local shadowing as gap fíllers do. Regarding planning and synchronisation aspeéis, the same rules apply to both sets of devices. However, gap fíllers must meet more stringent delay and transmission power requirements since they are typically not located at the fringe of the coverage área and their signal can more easily interfere with the signal of the main transmitter(s).
Figure 7.14 illustrates the rule that governs the use of gap fíllers within the coverage área of a single main transmitter, rated in this example at 25 kW effective radiated power (e.r.p.). In the case of DAB, this rule is closely related to the size of the guard interval (in Figure 7.14, Mode II is used with a 62 jxs guard interval). An omni-directional re-transmitting antenna is assumed at the gap fíller to cover the worst case situation. The propagation model of Recommendation ITU-R P. 1546 was used for this exercise, assuming fíat terrain with a roughness factor of Ah = 50 m. The domain of operation is under the curves. If the e.r.p. of the gap-filler exceeds the valúes shown by the curves at a given distance from the main transmitter and for a given gap-fíller, an unserved área starts to appear between the main transmitter and the gap-filler due to the presence of the destructive echo in this área.


Figure 7.14 Domain of operation of a gap-filler as a function of distance, HAAT and e.r.p.
In some cases, an alternative to gap filling would be to increase the power and/or the antenna height of the main transmitter. Apart from the cost implications, this would also increase interference to co-channel services in other áreas, and thus, limit imple-mentation of such a high-power DAB service, or reduce spectrum reuse efficiency. The use of gap filling transmitters therefore contributes to spectrum conservation.
Gap fillers do not have to be exactly time synchronised to the other transmitters in the SFN. The gap filler simply amplifies the received DAB signal. To achieve good performance the received signal may be downconverted to an IF or even baseband for signal conditioning before it is upconverted again for final amplification and filtering. De- and re-modulation for signal improvement are not done in gap fillers because the processing delay inherent in the DAB system means that the retransmitted signal would lie outside the guard interval irrespective of the DAB mode chosen. When a gap fíller is installed, it must be ensured that the transmit and receive antennas are sufficiently decoupled to avoid unwanted feedback and blocking effects. In general, the necessary isolation between receiving and transmitting antennas determines the upper limit for the amplification gain. At higher frequencies such as L-Band, large shadowing buildings in urban áreas can provide some of the necessary isolation.
In áreas where several DAB blocks are in use, the input stage of the gap filler must be block selective to guarantee that only the wanted signal is amplified. It is also possible that the gap filler could be used to amplify a number of blocks to cover the same gap. Proper filtering will be needed to avoid intermodulation. It should also be noted that in áreas with gap fillers in a DAB network, geographical position



estimation of the receiver using the Til feature wül not be possible for two reasons: fírst, because the transmitter time delay signalled in FIG 0/22 (TU field) is only valid for directly received signáis, and not for signáis which suffer from the additional delay due to the signal processing in a gap filler; and secondly, if the signal from the main transmitter and one or more gap fillers is simultaneously received, distinction of the different signal sources will no longer be possible since they will all have the same Til identifícation. (Gap fillers cannot change the TU information of the signal.) For more information on TU in DAB see section 7.4.7.
7.4.7 Application of the TU Feature in SFNs
DAB allows identification of individual transmitters in an SFN with the TU feature. The Til signal is transmitted every other nuil symbol to allow the receiver to perform channel state analysis in nuil symbols without the TU signal. The TU signal consists of a certain number of pairs of adjacent carriers of an OFDM symbol and the actual pattern of the carrier pairs identifies the individual transmitter.
The identification of each transmitter is given by two parameters: the pattern and comb number, also called the main and sub-identifier of a transmitter. FIG 0/22 in the FIC of the DAB signal describes a set of parameters, the Til field, which contains all information necessary for the unique description of a transmitter. These param¬eters are transmitter identifiers, geographical location of the transmitter and the time offset of the transmitter (see section 7.6.3).
The main identifier is used to describe a cluster of transmitters in a certain región and each transmitter within a cluster has its own sub-identifier. Table 7.2 gives the number of possible main- and sub-identifíers as a function of the DAB mode.
Each comb number identifies a number of carrier pairs of which only half is used in a Til symbol. Which of the carrier pairs are used is determined by the associated pattern number. Since each comb number identifies a unique set of carrier pairs (i.e. each carrier pair is only used by a specific comb), the DAB receiver can simultan¬eously identify the signáis of all transmitters with the same main identifier that is pattern number. To distinguish between transmitters with different main identifiers, the sub-identifier must be chosen carefully to avoid ambiguities. The exact relation-ship between comb and pattern is given in references [EN 300 401] and [TR 101 497].
The TU feature of DAB allows the receiver to calcúlate its position if signáis from at least three transmitters are received and FIG 0/22 (Til field) is signalled in the FIC [Layer, 1998]. The knowledge of the receiver position can be used for intelligent change of frequency when leaving the coverage área of the network. It can also aid
Table 7.2 Til parameters for different DAB modas
Mode
Number of Main identifiers (= diff. patterns)
Number of Sub-identifiers (= diff. combs)
Number of Carrier Pairs Used per Comb

1, IV, II III
70 6
24
24
4of 8 2of4




the automatic selection of information, for example only information relevant for the current región is displayed.
7.5 DAB Transmitiere 7.5.1 General Aspects
Figure 7.15 shows the block diagram of a DAB transmitter. Each transmitter consists of a number of functional blocks which will now be explained. The ETI output signal from the ensemble multiplexer is delivered to the transmitter site via the DAB distribution network. At the input of the transmitter the signal is buffered and a precise delay is inserted to synchronise the SFN in time. After COFDM encoding the baseband output signal of the COFDM encoder can be subjected to further signal processing for non-linear pre-distortion or crest factor manipulation before it is converted from digital to analogue. After conversión to the analogue domain the signal is upconverted to the desired final radio frequency. Finally the RF signal is amplified and filtered to fulfil the relevant spectrum masks before it is radiated.
7.5.2 Signal Processing Blocks of a COFDM Modulator
COFDM modulators usually contain not just the puré DAB signal processing part but also an input stage to process the different variants of the ETI signal and to insert the required signal delay. The output signal of the modulator is either the DIQ (Digital In-phase and Quadrature) baseband signal according to [EN 300 798] or an RF signal at a convenient IF or RF if an I/Q modulator is included. The signal processing blocks of a COFDM modulator are shown in Figure 7.16 and described in the following paragraphs.