Field trials with a high–power VHF single frequency network ...Field trials with a high–power...

36 EBU Technical Review Autumn 1994Maddocks et al.

Field trials with a high–power VHFsingle frequency network for DABMeasurement techniques and networkperformanceM.C.D. Maddocks

I.R. Pullen

J.A. Green

(BBC)

1. Introduction

FM sound broadcasting in Band II was designed asa high–quality system and can provide excellentsound quality. The advent of digital formats suchas CD has created a demand for uniformly high au-dio quality from radio. Also, listeners’ require-ments of a radio system have changed. People nowdemand high–quality and rugged audio reception(in stereo) in vehicles and on portable radios. Toprovide high–quality stereo reception to allreceivers, a Digital Audio Broadcasting (DAB)system capable of reliable reception in vehiclesand on portable receivers has been developed bythe Eureka 147 Project.

Conscious of the need to design a system that couldbe used in many different countries and situations,the Eureka DAB system was designed to work forboth terrestrial and satellite broadcasting at a rangeof frequencies. This is ensured by providing threemodes:

Mode 1, designed for wide–area, terrestrial broad-casting at VHF frequencies. Such networks arelikely to be extensively used in Europe.

The advent of digital formats such asCD has created demand for uniformlyhigh audio quality from radio, even invehicles and for portable reception.The Eureka–147 DAB system hasbeen designed specifically to meetthese demands and the BBC isundertaking a major experiment to testthe system and to gather data whichwill allow efficient planning of itstransmitter network. As a basis forthese tests, a network of four 1 kWe.r.p., VHF transmitters has beeninstalled to cover the London area inEngland.The results show wide–area coveragefrom the transmitter network which isin reasonable agreement withcomputer predictions. The results alsoprovide quantitative values which canbe used for coverage prediction andfor international coordination ofservices. Finally, the performance ofthe system demonstrates a number ofthe benefits of the Eureka DAB systemfor mobile and portable reception.

Original language: EnglishManuscript received 7/10/1994.

The DAB logo has been registeredby a member of theEureka 147 – DAB consortium.

37EBU Technical Review Autumn 1994Maddocks et al.

Mode 2, designed for local–area, terrestrial broad-casting at frequencies up to 1.5 GHz. Such net-works may be used in both Europe and countriesin other geographical regions, such as Canada,where the new 1.5 GHz (L–band) allocation isavailable.

Mode 3, designed for satellite broadcasting at fre-quencies up to 3 GHz. This may be particularly at-tractive in large countries, and for continental cov-erage.

In the United Kingdom, a VHF implementation forterrestrial broadcasting of DAB is preferred for thefollowing reasons:

– A national single frequency network (SFN) canbe economically implemented at VHF withlarge transmitter spacings and using existingbroadcasting transmitter sites.

– The propagation of signals, over terrain typical-ly found in the UK and into buildings, is gener-ally better at VHF than L–band.

– L–band is unlikely to be allocated in the UK un-til 2007. Even then it may be difficult to makeefficient use of the limited spectrum for bothsatellite and terrestrial DAB in the same cover-age area.

The BBC is undertaking a major experiment to testthe Eureka DAB system and to generate data to al-low efficient planning of its transmitter network.The article describes this experimental pro-gramme and the rationale behind it.

2. Planning for digitalsystems, mobile andportable receivers

Eureka DAB is explicitly designed to provide a re-liable service to mobile and portable receivers.This introduces a number of issues which must beaddressed in the system design and coverage plan-ning.

– Generally, digital modulation systems haveabrupt failure characteristics. This means thatthe received audio signal can change from highquality to total failure with only a small changein input signal level to the receiver.

– Mobile and portable receivers use omni–direc-tional antennas at a low height. Consequentlythe path from the transmitter to the antenna israrely unobstructed and the receiver obtains no

discrimination against multipath from other ob-jects.

– Portable receivers are usually used in buildings;this means that the additional loss and variabili-ty in signal level within buildings must be quan-tified.

3. Eureka DAB coding andmodulation system

Eureka DAB is based upon a coded orthogonal fre-quency–division multiplex (COFDM) modulationsystem [1]. COFDM modulation uses a large num-ber of RF carriers each of which is QPSK modu-lated at a relatively slow symbol rate. The carriersare spaced in frequency such that, as each carrieris demodulated, there is no interference from datacarried on adjacent carriers, ie they are orthogonal-ly spaced. The digital audio signals are protectedwith convolutional coding and the resulting data isdistributed across all the carriers in the RF band.Consecutive data samples are also separated intime. As a result, the audio data can still be recov-ered at the receiver even if some of the carriers cannot be demodulated owing to anomalous propaga-tion effects (such as multipath).

In the Eureka DAB system the problem of the swiftfailure characteristic of digital systems is ad-dressed by the introduction of unequal errorprotection of the audio signal. This process is de-scribed in more detail in [2].

However, two additional features are needed toimprove the performance of the system in multi-path environments:

– A wide bandwidth (significantly greater thanthe few hundred kilohertz currently used for FMbroadcasting) is needed to reduce the effects ofthe short–delay multipath that cause flat fadingproblems. Effectively this introduces frequencydiversity into the system; and a wide enoughbandwidth is needed to ensure sufficient decor-relation between the different components forthe diversity to be effective. To retain spectralefficiency with this bandwidth, a number ofprogrammes are bundled together.

– At the transmitter, each QPSK symbol is trans-mitted for longer than necessary to fulfil the re-quirements for orthogonal frequency spacing ofthe carriers. The extra time is known as theguard interval. At the receiver, an appropriateportion of this extended symbol (chosen to re-store the orthogonal frequency relationship) isdemodulated. The result is that echoes with a


delay of less than the guard interval do not pro-duce inter–symbol interference. The power inthis multipath signal can then be used construc-tively to aid demodulation.

An extra advantage of this system is that simulta-neous reception of the same information frommore than one transmitter is possible. The delayedsignals from the more distant transmitters will sim-ply look like multipath. This can be used to providemore reliable coverage of an area as it introducesspatial diversity. It also allows all the transmittersin a network to use the same frequency – thus pro-viding excellent spectrum efficiency. This mode ofoperation is known as the single frequency net-work (SFN).

4. Experimental work

A new system with these additional features posesnew challenges for efficient frequency and servicearea planning. For example, the SFN concept al-lows a choice to be made between serving an area(be it a city or a country) with a small number ofhigh–power transmitters or a larger number oflow–power transmitters. Or, the rugged digitalsystem allows much lower transmitter powers tobe used – but reliable coverage now depends on be-ing able to quantify and predict the amount of sig-nal variation which would be encountered by mo-bile or portable receivers.

For this reason the BBC has installed a network offour 1 kW, VHF, Band III, DAB transmittersaround London. With a population of around 10million people (18% of the UK population) this

provides a suitable test–bed for the system. Thepurpose of the experiment is to extensively test theDAB system in a realistic environment and gathercoverage data to allow more accurate planning ofthe BBC’s proposed transmitter network. It alsoprovides a platform for experiments on compo-nents of a DAB transmitter network and on receiv-er implementation.

Field–strength predictions and measurements arenormally made for 50% (median) location values.If the median field–strength in a particular area isequal to the minimum value for an acceptable ser-vice, the area is deemed to be served (assuming nointerference or other effects need to be consid-ered). In the case of an analogue system there willstill be a service to considerably more than 50% oflocations, but with reduced quality. For a digitalservice such as DAB, however, this would not bethe case. The transition from perfect quality to au-dio muting will occur over a field–strength rangeof relatively few decibels, depending on the sys-tem characteristics. In view of the rapid degrada-tion in digital systems near the failure point, it isnecessary to provide an adequate field–strength ina high percentage of locations. A figure of 99% hasbeen suggested for mobile reception. To achievethis, the median field–strength must be increasedby a suitable location correction factor (50–99%correction factor in this case).

The value of the correction factor is important inplanning a DAB service. In general, a small valueis desirable since it implies a lower median field–strength, and thus lower transmitter power. SinceDAB uses a signal of much wider bandwidth thanFM, there is expected to be a degree of frequencydiversity. This has the effect of reducing the varia-tions in the field–strength due to terrain effects,multipath, etc. Consequently the location correc-tion factor is reduced. The extent of this reductionwas the first object of the BBC study.

If an area is simultaneously served by severaltransmitters in a SFN, there is expected to be adegree of spatial diversity, resulting in a furtherreduction in signal variation. The second purposeof this work has therefore been to quantify theimprovement.

Finally, the amount of reduction in signal level forreception in buildings has been examined.

4.1. Transmitting system

For this experiment, four DAB transmitting sta-tions were installed. Sites already used for broad-casting conventional television and radio pro-grammes were selected, because these allowed the

Figure 1Locations of theLondon DABtransmitters.


experimental signal to be radiated from an antennamounted at a representative height above theground. The locations of the sites selected areshown in Fig. 1 and were:

Crystal Palace The main transmitter for UHF tele-vision transmissions to London, located in southLondon.

Wrotham The main transmitter for VHF/FM Radiotransmissions to London, located to the east ofLondon.

Reigate A transmitter situated on a ridge of hills tothe south of London and used to transmit televisionprogrammes at UHF to towns to the south of Lon-don.

Alexandra Palace A transmitter situated to thenorth of London and originally used to transmittelevision programmes at VHF to London.

Provision was made to transmit an e.r.p. of 1 kWfrom each transmitting site. In addition, the CrystalPalace transmitter was specified to transmit ane.r.p. of 10 kW, if required. The radiation patternof the Crystal Palace DAB transmission was ar-ranged to be approximately omni–directional. Theother three transmitters have patterns which directmost of the power towards London. This allowsthe SFN concept to be tested.

For the first stage in the experiment the DAB sig-nal was generated at the Crystal Palace transmitterand distributed at UHF to the three surroundingsites. This allowed simple transmitters consistingonly of frequency translation and amplifyingequipment to be installed at the other sites. As thenetwork moves towards a pilot DAB service forLondon, more sophisticated distribution methodsare being installed.

4.2. Measuring system

For both mobile and portable measurements aDAB receiver, signal level measuring equipmentand a computer were installed in an estate car. Thesurvey vehicle (Fig. 2), which was speciallyequipped for DAB measurement, had a low roofline so as to be representative of normal car recep-tion, i.e. approximately 1.5 m. A Band III receiv-ing antenna was fitted in the centre of the roof togive a response as close as possible to omni–direc-tional. Additional antennas were fitted to providesignals to the positioning system and mobile tele-phone.

The system was designed to allow measurementsto be made of position, received signal strengthand DAB system performance, up to ten times asecond whilst the vehicle is on the move. This datais analysed later. The approach allows the largeamounts of data gathered to be processed efficient-ly and effectively.

4.2.1. Hardware

The vehicle is fitted with a third generation DABreceiver and a Rhode & Schwarz ESVB testreceiver to measure the field–strength. Audiomaterial from the DAB receiver is played throughthe car radio system to allow subjective assessmentof the audio quality to be made by the engineers.The following additional information is recordedon a portable PC:

– Violations of the audio scale factor cyclic re-dundancy checksum (CRC) and occurrences ofmuting are recorded as an objective measure ofaudio performance. This is important for tworeasons; firstly, it takes into account the com-bination of degradations due to low signalstrength, long–delay (and hence interfering)DAB signals, and non–DAB interference suchas man–made noise. Secondly, it allows the per-formance of the signal to be measured withoutthe need to transmit specific test data – thus theaudio programme can continue to be broadcast.

– DataTrakTM and GPS positioning systems arefitted which give positional information toaround 50 m accuracy. However, experimentalexperience has shown that the positional in-formation becomes unreliable when the vehicleis in dense forests and some dense urban areas.

Figure 2Interior of the DAB

survey vehicle.


Band IIIDAB

antenna

PC

Phonediplexer

Positioningsystem

Mobilephone

antenna

ESVBTest receiver

DABreceiver

Spectrumanalyzer

Signal level

Audio failures

Position

Amplifierandfilter

Impulseresponse

Figure 3Mobile measuringequipment.

In addition, an oscilloscope and spectrum analyserare provided to assist the engineer making the mea-surements. The former allows the impulse re-sponse (derived from the DAB receiver) to be dis-played, thus showing the number and relativemagnitude of the signals being received. The latterallows continuous–wave (CW) interferers andman–made noise to be identified. The resultingblock diagram for mobile measurements is shownin Fig. 3.

Considerable care has been taken to screen the ex-perimental hardware as it has been found that radi-ation from this equipment (particularly the DABreceiver) can significantly limit the performanceof the system. It is anticipated that this problemwill be reduced in later versions of DAB receiverswhich incorporate customised ICs.

In cases where measurements of portable recep-tion quality were to be made in houses, the antennawas taken inside, but the rest of the equipment wasleft in the vehicle. For measurements in buildingstwo types of receiving antenna were used. The firstconsisted of a folded–dipole antenna and was cho-sen to provide a well–matched antenna for reliablemeasurement of signal strength, Fig. 4. The secondwas a whip antenna installed on a box of approxi-mately the same size as a conventional radio re-ceiver. This was used for making measurements ofexpected signal quality in buildings, Fig. 5.

In both cases a large number of measurements ineach area were made so that valid signal statisticscould be calculated – even for relatively small per-centages of locations.

Figure 4Field–strengthmeasurement in ahouse.

Figure 5Mock portable DABreceiver used toestablish subjectiverequirements forcoverage.


4.2.2. Logging and analysis software

In order to meet the survey data–acquisitionrequirements of the high–power DAB field trials,a logging program has been developed. The pro-gram runs under Windows 3.1 on an IBM 386 (orhigher) computer and presents an easy to use,graphical interface to the operator. At the heart ofthe program is a fully configurable system for de-fining the devices and interfaces used to gather thedata. A number of interface cards are supported,including RS232, GP–IB and DMA (parallel inter-face) cards. So long as it has an interface that issupported by the data logging software, a suitablecommunication link between computer and mea-surement instrument can be configured. There isno need to write specific software to communicatewith a particular type of measurement device. Thisenables the software to be used for a large numberof different data gathering functions.

Once a configuration has been established, it maybe saved together with the default information thatis to be stored in the data file. This information in-cludes the parameters of the system for which thedata was gathered, the date and time, and the iden-tity of the operators.

In addition to the measured data calibration in-formation, survey and project–related details arerecorded. The information is stored in a file formatwhich is directly compatible with the suite of anal-ysis programmes.

During data acquisition, the data is displayed to theuser as it is gathered, giving a reliable indication ofthe on–going performance of the system (Fig. 6).The same display format is available to displaypreviously–logged data files for comparison

Software has also been written to perform analysisof survey data gathered during field trials. The datafiles are processed by a suite of programs whichhave been developed in “C” and written so thateach program performs a separate operation in theanalysis procedure. A full audit trail is maintainedof the processes performed on the data.

A series of programs may be run on a survey datafile to produce the desired analysis. Each individu-al program performs a simple function on the datasuch as sorting it, finding various statistics, dis-playing it, etc. The required analysis is performedby assembling these routines in the required order.This provides a very flexible analysis tool. The re-sults can be output to wordprocessor or spread-sheet applications. Each program has a wide vari-

ety of options so that individual programs may beeasily tailored to do a specific task in the analysisprocedure.

4.3. Experimental methodology

4.3.1 Survey areas

The first priority in the experimental work was tomeasure the signal level and system performancewith individual transmitters (each at 1 kW e.r.p.).Only then could the effects of multiple transmittersbe examined. This would simultaneously provideinformation about the performance of the systemand allow comparisons between the measured andpredicted signal levels in an area.

The area expected to be covered by the transmitternetwork is large. To obtain reliable statistics for thesignal level and audio quality available at 99 % oflocations in a small area requires a large number ofmeasurements in each area. It is clearly impractical(and unnecessary) to survey every area in such de-tail. Therefore predictions of the expected cover-age were obtained and used to target “interesting”areas. In general, these were areas which were ex-pected to be most marginal. However, in the earlystages of the experimental work, an exception wasmade and relatively wide areas were surveyed inlesser density. This was required to identify anysystem problems which might occur in areaswhich were expected to be served – and hencewould not otherwise be surveyed.

Figure 6DAB data–logging

software.


A version of the BBC field–strength predictionmethod which was developed for planning con-ventional, analogue television and FM radio cov-erage was used [3]. However, modifications havebeen made to allow for the different characteristicsof the digital signal, and the low gain and height ofthe receiving antenna installation [4]. The pre-dicted coverage for two of the individual transmit-ters and the total network is shown in Fig. 7. Thesepredictions are presented in terms of the expectedpercentage of locations served in each area. To thewest of London, relatively flat ground results in aneven reduction in signal level with distance fromthe transmitter. However to the south of Londonthere is a region with many small hills and valleyswhere very localised changes of coverage are ex-pected. A wide range of types of ground clutter arealso encountered, as the centre of London is adense urban area, whereas further out there aresuburban and rural areas.

Measurements were also made in domestic houses.The purpose of this work was to assess the buildingpenetration loss in domestic houses and to obtaina first estimate of the coverage level required bylisteners. In this study, the distribution of houseswas limited to a selection of those which BBC staffwere prepared to make available within the pre-dicted service area .

4.3.2 Measurement and analysistechniques

a) Mobile work

Having targeted areas of particular interest, theywere surveyed in detail. For this work, the area wasdivided into 500 m by 500 m squares and measure-

c) Complete network.

a) Crystal Palace transmitter only.

b) Reigate transmitter only.

Figure 7Predicted coverage.


ments were made in a number of these squares inan area. The use of these squares allowed easycomparison of measured and predicted results, asthe prediction algorithms currently have a maxi-mum resolution of 500 m.

A number of forms of analysis were performed:

– The mean and standard deviation of the mea-sured field–strengths measured in each squarewere found. The predicted field–strength wasalso calculated.

– The measured field–strength and the receiverstatus were compared to find the minimumfield–strength for correct operation of the re-ceiver.

– The objective measure of audio quality derivedfrom the DAB receiver was compared with thesubjective result recorded by the engineer in thevehicle.

b) Portable work

In this work measurements of signal level weremade in the 500 m by 500 m square in which thehouse was located, immediately outside the houseand in various ground–floor rooms inside thehouse. The results could then be analysed to findthe average building penetration loss and the varia-tion in signal level within a house.

4.3.3. Equipment set–up

The logging system was set to record:

Time from its internal clock

National Grid from the positioning systemReference (NGR)1

Received power from the ESVB receiver

Audio scale factor from the DAB receiverCRC errors (a coarsemeasure of audio quality).

Readings were initially triggered by elapsed timeat the rate of one per second. The ESVB receiverwas set for an integration time of 10 ms. Thismeant that each reading would be averaged overabout 10 DAB symbols so that variations in signallevel within a symbol period (and across a nullsymbol) was averaged out. The receiver filter wasset to 1.5 MHz in line with the bandwidth of a 3rd–

1. The National Grid Reference serves to define anyposition in the United Kingdom in relation to a system oflinear coordinates in a “flat–Earth” projection.

generation DAB signal, and the centre frequencywas set to 226.25 MHz, the centre frequency of theexperimental transmissions.

After some initial experiments, the arrangementwas re–configured to trigger each 96 ms COFDMframe. Measurements were then made more fre-quently, allowing greater detail of the fast–fadingstatistics to be recorded. The integration time ofthe ESVB was correspondingly reduced. Synchro-nisation of the measurement time to the COFDMframe meant that measurements of signal strengthduring the null symbol could be avoided.

4.4. Future improvements

A number of improvements to the experimentaltechniques and equipment can be suggested as a re-sult of our recent experiences. The simplestchanges relate to a reduction or the elimination ofthe amount of measuring equipment requiringmains power. A move to an entirely 12–V poweredarrangement is envisaged in the near future. How-ever there are two somewhat longer–term changeswhich are being considered.

Measurements of the total received signal strengthcan be made satisfactorily using a conventionalmeasuring receiver or a specialist DAB receiver.The only detailed requirement for DAB is thatfield–strength measurements should not be madeduring the null symbol. However, measurementsof the signal strength from an individual transmit-ter will be required to assess the effect of modify-ing an SFN. That is, it is highly desirable to mea-sure the field–strength of one transmitter in an SFNon its own. This would be used in an operationalenvironment when the benefit of adding a fill–intransmitter was being considered.

This measurement can be achieved in a number ofways; one such method would involve using theDAB transmitter identification information (TII).The power in the comb of carriers radiated in thenull symbol could be calculated. The delay of thesignal can be found from the difference in phasebetween adjacent carriers in the comb and thetransmitter can also be uniquely identified. Theonly problem with this method is that measure-ments could only be made about 5 times per second(for a Mode 1 signal). However, for most pur-poses, this may be satisfactory when used in con-junction with a measure of total signal strength,made more frequently.

The second change being considered is a computerdisplay showing DAB performance overlaid on amap. This would greatly improve the targeting ofoperational survey work.


5. Results

5.1. Calibration

Two calibration issues were considered. Firstly,confidence was needed that the calibration factorarising from the antenna and amplifier gain, feederloss and voltage–to–power conversion had beencorrectly measured and calculated. To do this, mea-surements at a test point with a line–of–sight pathto a transmitter were made and compared with pre-dictions.

Secondly, confidence was needed about the repeat-ability of field–strength measurements. For thispurpose, three sets of measurements were madeover a short test route with the vehicle travelling atdifferent speeds and in a different direction aroundthe route. The results for each run were comparedand showed differences of significantly less than1 dB.

In each case the results provided confidence that re-liable and repeatable results could be obtained.

5.2. Coverage area of transmitters

5.2.1. Coverage of individualtransmitters

The coverage of each transmitter was measured intwo ways. First, the subjective quality of thereceived audio signal was noted. Three categorieswere used to record this data; areas where the audiosignal was unimpaired, areas where the signal wasunintelligible and the areas in between. Second, theobjective results of the receiver status were ana-lysed as the number of CRC errors and muting

events within each 500 m by 500 m square. Thesquares were classified into one of three categories:

Unimpaired: Less than 1% of samples withCRC errors and no mutes within square

Marginal: More than 1% of samples withCRC errors, but no mutes

Impaired: One or more muting events in square

In cases where fewer than 200 samples were re-corded, a small modification was made. A singleisolated error may not be audible; for this reason asquare is only deemed to be marginal if two or moreerrors were logged. Mutes, on the other hand, al-ways last for at least one second. Thus the minimumcriteria for a square to be deemed impaired was justone mute. It should be noted that this is a very oner-ous criteria as a single mute can be caused by a smalltunnel or bridge, or a high level of man–made noisefrom a single building. Fig. 8 shows that goodagreement is obtained between the subjective andobjective measurements.

The results of the objective measurement of audioquality for two of the individual transmitters areshown in Figs. 9 and 10. Whilst the whole area hasnot been surveyed, comparisons with predictions inFigs. 7a and 7b show results which are similar tothose predicted. Some trends can be identified in thedifferences. Coverage in rural areas was generallyfound to be better than predicted, and coverage indense urban areas slightly worse. This is probablybecause the BBC prediction programme has beenoptimised for predictions in suburban areas. Thesystem performance in some urban areas was alsofound to be affected by man–made noise. At asmall percentage of locations, high levels were ob-served.

a) Subjective coverage b) Objective coverage

Figure 8Comparison ofsubjective andobjectivemeasurements ofmobile coverage.


Figure 9Measured coverage

of the CrystalPalace transmitter.

Figure 10Measured coverage

of the Reigatetransmitter.


Figure 11Comparison ofmeasured andpredictedfield–strengths.

Predicted field–strength (dB�V/m)

Measuredfield–strength

(dB�V/m)

90

80

70

60

50

40

30

20

10

0

80706050403020100 90

Space in this article does not permit a detailedcomparison between the predicted and measuredresults. However, the general conclusion is that theDAB signal covers a somewhat larger area thanpredicted. Fig. 11 compares the median values ofthe measured and predicted field–strengths. Thisshows a trend at low field–strengths of the predic-tion programme under–estimating the field–strength. The difference probably explains the dif-ference in coverage areas.

5.2.2. Coverage of the full network

Initial measurements have targeted areas in whichthe coverage was expected to be marginal. Find-ings so far have shown the large expected benefitin system performance arising from the simulta-neous reception of several signals.

An example of the effect is shown in Figs. 12, 13and 14. Fig. 12 shows an area in the south west ofthe Greater London area which is on the edge of the

Figure 12Terrain in the regionof Woking and Leatherhead(vertical scale exagerated).


Figure 13Predicted coverage.



c) Crystal Palace and Reigate transmitters together.


Figure 14Measured coverage.



c) Crystal Palace and Reigate transmitters together.


service area from both the Crystal Palace and Re-igate transmitters. The area is also interesting as itcontains a relatively flat area to the north and amore hilly ridge of land in the south of the area.The predicted coverage of the area from the indi-vidual transmitters and for the whole network isshown in Fig. 13. These show that a significantbenefit from “network gain” can be expected inthis area but, even so, the area is only expected tobe marginally served.

The area was also measured in each of the threeconditions. The results are shown in Fig. 14. Theseshow a somewhat better measured coverage than

that which was predicted. They also show that theexpected benefit of anticipated network gain isrealised in practice. As a result, the only area whichwas not found to be covered was a small valley inthe hilly area. This is actually a well documenteddeficiency for analogue signals and is served by atransmitter to the south. Detailed coverage resultsfor the whole network have been made and similar-ly show that the predicted coverage is being ob-tained.

5.2.3. Network gain

The transmitter and logging arrangement has al-lowed the fast–fading statistics and the correlation

Field–strength(dB�V/m)

Reading number

a) Not averaged.

b) Moving averaged.

80

60

40

20

0

1 101 201 301 401 501 601 701 801 901


80

60

40

20

0

1 101 201 301 401 501 601 701 801 901

Figure 15Field–strength

vs. distance(Epsom Downs

test route).



Percentage of locations for which field–strength is exceeded

Figure 16Cumulative distribution for thetest route shown in Fig. 15(Moving average data).

70

65

60

55

50

45

4050 70 90 951 5 10 30 99

between signals from different transmitters to beexplored in more detail. The correlation is, ofcourse, an important quantitative measure of thenetwork gain that can be expected.

To enable this the transmitters were put in a statewhere they were switched on and off eachCOFDM frame. The correlation between signalsreceived quasi–simultaneously from differenttransmitters could be measured. This effect is dem-onstrated in Figs. 15 and 16. Fig. 15a shows the re-ceived signal level from the Crystal Palace and Re-igate transmitters alone and the case when both areswitched on. These measurements were made onaround 100 m of road. The superposition of thefast–fading (local multipath) variation on the

Amount ofbuilding cover

(%)

Number ofareas in the

category

Weighted aver-age standard

deviation(dB)

100

90

80

70

60

50

40

30

20

10

occasional

0

151

215

167

144

126

148

141

105

150

181

509

170

4.3

4.1

4.1

4.0

3.7

4.2

4.1

3.8

4.0

4.0

4.2

3.9

Total 2217 4.1

slow–fading variation can be seen. As it is theslow–fading component of the signal variation ina local area which is particularly important in theconsideration of network gain, the fast–fadingcomponent was removed by calculating a movingaverage across a number of measurements. Theslow–fading component is shown in Fig. 15b. Itcan be seen that there is significant decorrelationbetween the signals from the two transmitters. Theeffect on the network gain at the 99% locations lev-el can be found by calculating the cumulative dis-tribution; this is shown in Fig. 16. At the 99% loca-tions point, Fig. 16 shows that the performance ofthe system with both transmitters operating isaround 6 dB better than the performance with onlythe Crystal Palace transmitter. This result is partic-ularly important as the Reigate transmitter is pro-ducing a median field–strength in the area whichis considerably lower than that from Crystal Pal-ace.

5.3. Local–area variations in signalstrength

The measured results were analysed to find the lo-cal–area variation in the slow–fading componentof the signal. For this purpose, measurements gath-ered in each 500 m by 500 m square over the areawere analysed to find the median signal levels andthe standard deviation of the variation. Sampleareas were checked to ensure that the signal dis-tribution was a reasonable approximation to lognormal.

The standard deviation of the local area variationsin signal level were averaged for different terraintypes. The information about terrain types was ob-tained from a clutter database of the area which

Table 1Standard deviation ofthe signal variation inlocal areas.


Figure 17Variation of

field–strength in a localarea.

Figure 18Difference in amount of

signal variation forwideband and narrowband

systems.

Number ofoccurrences

Standard deviation (dB)


Percentage of locations for which field–strength was exceeded

100

80

60

40

20

0

0 1 2 3 4 5 6 7 8 9 10

80

60

40

20

0.1 0.5 1 2 5 10 20 50 80 90 95 98 99

categorises each square by the percentage of thearea covered by buildings. This percentage pro-vides an approximate indication of rural, subur-ban, urban and dense urban areas. In averaging thestandard deviations, the results were weighted bythe number of samples recorded in each area. Theresults are summarised in Table 1.

The results for over 2200 areas showed an averagestandard deviation of 4.1 dB. The spread in the val-ues of standard deviation is shown in Fig. 17.

The results show (as expected) significantly lesssignal variation than occurs with narrowband sys-tems, for which a value of 9 dB is implicitly as-sumed in ITU–R Recommendation PN.370 [5].This confirms a result which has been measuredusing low–power transmitters in an earlier net-

work. In that experiment, measurements weremade in the same area, from the same transmitter,of the signal variation using a wideband and nar-rowband signal. The results, Fig. 18, showed stan-dard deviations of signal variation of 9 dB and5 dB for the narrowband and wideband signals re-spectively.

The second point to note from these results is thatthe standard deviation is slightly lower than hasbeen measured before [6]. The reasons for this arestill being studied in detail, but the most likely are:

– that the current measurements concentratemore on the slow–fading component of the sig-nal and consequently a longer power integra-tion period is used than before;

– a wider variety of terrain has been surveyed.


The second point is important, in that earlier ex-periments were confined to a relatively small area,much of which was located in relatively undulat-ing terrain. The effect of this is to increase theamount of signal variation in a small area. The cur-rent work explores a wider range of terrain types.

The third point to note from the results in Table 1is that the amount of ground clutter appears to haverelatively little effect on the amount of signal vari-ation. This is perhaps an unexpected result. Onceagain this area is still being studied, but it is pos-sible that any variations due to differences inground clutter are being swamped by differencesin the amount of terrain undulation.

5.4. System performance results

The purpose of this work was to determine the fail-ure field–strength for the DAB receiver. To dothis, the mobile measurements were analysed todetermine:

– the maximum signal strength in an area wherethe audio was not badly degraded;

– the minimum signal strength in an area wherethe audio was degraded.

The results showed a wide variation in field–strength with a mean failure field–strength of

around 40 dB�V/m. This result is slightly higherthan the value sometimes assumed for DAB plan-ning; however, the current measurement arrange-ment is known to suffer from higher levels of man–made noise than would be expected in a domesticarrangement. Considerable care has been taken inthe measuring vehicle to screen the relevant equip-ment; however, even so, a significant cause of theman–made noise is the 3rd–generation DABreceiver itself.

5.5. Measurements in buildings

5.5.1. Objective measurements

An important value for planning the coverage toportable receivers inside buildings is the buildingpenetration loss. That is, the difference betweenthe signal level inside the building and the levelthat would have been received in the same place ifthe building were not there. This is found by mea-suring the signal statistics outside the building, onthe side nearest the transmitter, and the signal sta-tistics inside. The difference in median values canthen be found.

39 building penetration losses have been mea-sured. The first 26 were measured using low–power transmitters and the results have beenreported in [7]. Most of these were typical domes-

Mr. M.C.D. Maddocks is a Senior Engineer in BBC Research and Development Department. He joinedthe BBC in 1982 and has worked on a range of RF projects covering digital modulation, antenna designand propagation. He is involved in various national and international groups, and is Chairman of theEureka 147–DAB Working Group responsible for RF, channel coding and modulation.

Mr. I.R. Pullen graduated from the University of Kent at Canter-bury in 1976 and gained a PhD from the same university in 1980.After working as a research fellow at the university, he joined theBBC Research and Development Department in 1984, where hehas been involved in work on NICAM, RDS and DAB. He is ac-tive in EBU Specialitst Group R4/Prediction and in ITU–R StudyGroup 10B.

Mr. J.A. Green joined the BBC in 1973 and has been with the Service Planning Section at BBC Researchand Development Department since 1978. He graduated from the Open University in 1992. He hasworked on a variety of research projects including the development of objective measurements for VHF/FM reception quality in vehicles, and the techniques and methodology of DAB coverage surveys.


tic houses located in suburban areas. During thehigh–power experiments a further 13 buildingshave been measured. These concentrated on build-ings in urban and dense urban areas.

The first set of measurements found an averagebuilding penetration loss to ground floor locationsof 7.9 dB. As expected, it was found that signal lev-els on upper floors in houses were higher. The mea-surements concentrating on houses and basementflats in built–up areas were expected to find a high-er building loss; however an average value of 8.3dB was measured. This was not significantly dif-ferent from the earlier measurements. On reflec-tion, the absence of change with degree of groundclutter is believable as the clutter loss will usuallyaffect both the signal level outside and inside thehouse. The spread of building penetration lossesmeasured is shown in Fig. 19.

The average standard deviation of the signal varia-tion, considering only the measurements taken onthe ground floors of houses, was found to bearound 4 dB.

5.5.2. Subjective requirements forreliability of coverage

One of the tasks identified earlier was to investi-gate the subjective coverage requirements for por-table reception in buildings. That is, to make somemeasurement of the percentage of locations withina building which must be covered before, on aver-age, a listener will find the service acceptable.Such a figure could then be incorporated into cov-erage criteria.

Investigating this subject is fraught with difficul-ties. As with all subjective testing, the problem ofjustifying terms such as what constitutes a serviceis difficult. The problem is exacerbated by the factthat not only are a number of different peoplesampled, but also the opinion of each of them issought in a different house. Finally, their listeninghabits, expectations and uses of radio receivers areoften different.

As a first step in investigating this issue, the opin-ion of the owner of the house was sought as to theminimum acceptable coverage in their house; thiswas done when making the penetration loss mea-surements. An attenuator was added in the signalpath between the receiving antenna and the receiv-er. This was adjusted until the house was justdeemed to be served. The percentage of locationsin the room for which the audio did not have scalefactor CRC errors was then measured.

The results suggest that around 80% of locationsshould be served within a room before it is deemedto be served. However, this percentage variedgreatly from house to house or person to person.Primarily, it appears that this is because there arerelatively few places where people are prepared toput receivers. If it works there – fine; if it does not,then it is less than acceptable.

The second important point was that, once a satis-factory location had been found, it was very impor-tant that it should stay satisfactory as peoplemoved around the room and other channel distur-bances occurred.

6. Discussion

The first results from the network of transmittersaround London have demonstrated a number of in-teresting properties of the Eureka DAB system.They have confirmed, over a large number of mea-surements made in many different areas, that thefrequency diversity which is inherent in the systemprovides a significant reduction in the amount ofsignal variation. This is an important property ofany digital system designed to serve mobile andportable receivers, as a very high level of availabil-ity for the system is required.

The results have also provided an important val-idation of the performance of the system. Goodaudio reception has been found over the area ex-pected from predictions and simulations of the sys-tem. This indicates that the system and the firstequipment built to the system specification is oper-ating as expected.

Detailed analysis of the measurements obtainedwhen several transmitters are received simulta-neously is still in progress, but the results so farshow significant improvements in coverage evenwhen there are relatively large differences in the

Figure 19Spread of building

penetration losses.

0

1

2

3

4

5

6

7

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Numberof

occurences

Loss (dB)

Mean value: 8 dBNo. of measurements: 39


median signal levels from the different transmit-ters. This supports earlier experiments which haveshown a reduction in the amount of signal variationwhen multiple signals are received and a conse-quent reduction in errors in the received datastream [8].

An interesting consequence of this work is that sat-isfactory reception has been found in most of theLondon area using relatively–low transmittere.r.p.s.

The results have provided data which is being usedto refine the algorithms and methods used for cov-erage prediction. This is an important input at atime when the BBC is planning the locations ande.r.p.s of the network of transmitters which may beused to serve the whole of the United Kingdomwith DAB services. Experience in the develop-ment of these techniques is showing that the exper-imental measurements can have a major impact onthe accuracy of the coverage prediction algo-rithms. Significant modifications to the techniquesnormally used for analogue broadcast signals arerequired.

However, there are still a number of important top-ics which are being investigated using the network.Examples are:

– Experiments are being performed to measurethe correlation between signals from differenttransmitters. This is an important measure toquantify the benefit derived from the spatialdiversity (which occurs when signals frommore than one transmitter are received simulta-neously).

– Different solutions for local–area broadcastingusing the system will be investigated.

– More accurate quantification of the magnitudeand effect of long–distance interference fromtransmitters is needed. Such work is, of course,time consuming as high levels of interferenceonly occur for a small percentage of time andlong measurement periods are required toachieve statistically significant results.

7. Conclusions

This article has described the first results fromfield tests of the Eureka DAB system. These testsare being conducted in and around London and usefour 1 kW, VHF transmitters operating as a singlefrequency network.

The results show wide–area coverage from thesetransmitters which is in reasonable agreement with

experimental predictions. This indicates that thecurrent equipment, built to the Eureka DAB speci-fication, is operating in the way that would be ex-pected from theoretical studies and simulation.

The results also provide quantitative values whichcan be used for coverage prediction and for in-ternational coordination of services. These valuesare similar to those measured in earlier work.

Finally, the performance of the system demon-strates a number of the benefits of the Eureka DABsystem for mobile and portable reception. In par-ticular, it shows the reliability that can be achievedwithout needing very high–power transmitters;this results from the frequency diversity which isinherent in the system and the spatial diversitywhich can be obtained if more than one transmitterradiates the same signal at the same frequency. Allthis has given the BBC the confidence to proceedwith plans towards implementing a UK nationalsingle–frequency DAB network for its nationalprogramme services.

Acknowledgements

The authors would like to thank their colleagues,at both BBC Research & Development and BBCTransmission, for their work in setting up the trans-mitter network and receiving equipment, and mak-ing and analyzing the measurements presentedhere. The authors would also like to thank the BBCfor permission to publish this article.

Bibliography

[1] Alard, M., Lassalle, R.: Principles of modu-lation and channel coding for digitalbroadcasting for mobile receiversEBU Review – Technical, No. 224, August1987, pp 168–190.

[2] Ratliff, P.A.: The Eureka–147 DAB System –The system for fixed, mobile and portablereceivers2nd International Symposium on Digital AudioBroadcasting, March 1994.

[3] Causebrook, J.H. et al.: Computer predic-tion of field–strength: A manual on meth-ods developed by the BBC for the LF, MF,VHF and UHF bandsBBC Research Department.

[4] Lee, M.B.R.: Planning methods for a na-tional single frequency network for DABProceedings of 8th International Conferenceon Antennas and Propagation, April 1993,Conference Publication Number 370, pp940–947.


[5] ITU–R Recommendation PN.370: VHF andUHF propagation curves for the frequencyrange from 30 MHz to 1000 MHz. Broad-casting service

[6] Maddocks, M.C.D., Pullen, I.R.: Digital Au-dio Broadcasting: Comparison of cover-age at different frequencies and with dif-ferent bandwidthsBBC Research Department Report No. BBCRD 1993/11.

[7] Green, J.A.: Building penetration loss mea-surements for DAB signals at 211 MHzBBC Research Department Report No. BBCRD 1992/14.

[8] Maddocks, M.C.D., Pullen, I.R. and Lee,M.B.R.: Experimental work to assist inplanning a DAB single–frequency networkIEE Colloquium digest No. 1993/043, pp9/1–9/8.

The IAB welcomes its first students

Located in Montreux, Switzerland, the International Academy of Broadcasting has been createdwith the aim of offering interdisciplinary postgraduate studies in the art and science of radio and televi-sion broadcasting. Now, after one year of intense preparations, the Academy has opened its doors tothe first generation of students.

Arriving from different parts of the world (Algeria, Ghana, Hungary, Iceland, Romania, Sweden, Switzer-land, Ukraine and ex–Yugoslavia), and with different university backgrounds, the IAB students gath-ered on 23 September 1994 to celebrate the official opening of their school. Also present at the OpeningCeremony were IAB Council members, professors and patrons of the IAB, representatives of otherschools, numerous guests from the EBU Member organizations, the EBU Headquarters and the Mon-treux and Vevey municipalities, along with distinguished members of the local community.

The President of the IAB Council, Dr. George T. Waters, welcomed the students and guests, and ex-pressed his happiness to see “history in the making” ... “a dream becoming true”. Monsieur ErnestGuibert, representing the Montreux municipality, stressed the im-portance of establishing such a prominent academic institution inMontreux. Commenting that the city is already well known for its me-dia events and its high–quality educational institutions, he ex-pressed his personal belief that the IAB would successfully grow onthat fertile ground.

The highlight of the Opening Ceremony was the appearance of thePresident of the EBU, and Patron of the IAB, Professor AlbertScharf. He wished the Academy and its students every success,and stressed the importance of this new educational institution tothe future of the broadcasting industry. Professor Scharf closed hisinspiring address by declaring the International Academy of Broad-casting open.

Another IAB Patron, Sir Peter Ustinov, was unfortunately unable toattend the Opening Ceremony. However, he was neverthelesspresent in a way which is most appropriate for a broadcasting school– through a video tape recording.

Substantial support for the IAB has been forthcoming from thebroadcast industry and the Academy has been especially fortunateto be given a fully–equipped earth station by Scientific Atlanta. Tak-ing advantage of the presence of representatives of the industry atthe IAB Seminar on Digital Video Broadcasting, a reception was or-ganized to mark the official hand–over of this facility by Mr. D. Ozley(Scientific Atlanta).

Presentation of the IAB earth station(l–r: Prof. A. Todorovic, Mr. D. Ozley, Dr. G.T. Waters)

Field trials with a high–power VHF single frequency network ...Field trials with a high–power...

Documents

Transcript of Field trials with a high–power VHF single frequency network ...Field trials with a high–power...