LOCAL-AREA FIELD STRENGTH VARIATION MEASUREMENTS OF THE DIGITAL SIGNAL (COFDM 8K) IN SUBURBAN ENVIRONMENTS

A. Arrinda, M. Mª Vélez, P. Angueira, D. de la Vega, J. L. Ordiales UNIVERSITY OF THE BASQUE COUNTRY

ETSII y IT de Bilbao Alda. Urkijo s/n 48013 BILBAO, Spain ABSTRACT

This paper presents the results of a field First, a general overview of the experimental measurement campaign at the experimental network is given. A description of measurement COFDM 8K digital terrestrial television equipment and techniques will follow. Finally, network developed by the partners of the results and discussion will be presented. VIDITER project in Madrid (Spain). This article focuses on the field strength spatial EXPERIMENTAL NETWORK variation distribution inside small coverage areas. Due to the fact that distribution Measurements presented here have been taken parameters strongly depend on the survey inside the coverage area of the digital terrestrial area type, during this first stage the only television experimental network located in Madrid environments taken into account have been (Spain). This network was designed for test those traditionally considered as suburban. purposes and allows broadcasting according to the European DVB-T standards [1]. This work was the INTRODUCTION result of the co-operative research and development project VIDITER, led by the Spanish Government, Plans have been set up for the technical launch of the company RETEVISION, digital terrestrial television (DVB-T) services in Spanish broadcast equipment manufacturers and some European countries as Sweden, Finland, the University of the Basque Country. The design of Ireland and Spain. The target on-the-air deadlines this network has included all the aspects to be for this launch are the last quarter of 1999 and the considered in the near future when DVB-T first months of year 2000. Up to now, and excluding production, transport and broadcast networks will United Kingdom where the selected mode is the be fully operational. DVB-T 2K, the only remarkable experiences with the 8K system have been achieved by means of Video and audio sources as well as compression experimental networks. Some of the most equipment are located at the RETEVISION interesting aspects of all these tests have turned out laboratories. MPEG Transport Stream is distributed to be the different field measurement campaigns. by means of different networks to the Results from these activities should become a sites. This research centre acts as a program reference for nation-wide network planning. production centre. Two main sites constitute the broadcast infrastructure: the first one is located in The main goal of this paper is to show the results of the city of Madrid and the second one is on the a field measurement campaign aimed at surveying Navacerrada mountain range, 50 Km away from the the field strength spatial variation over small areas. city at 2256 meters above sea level. The Estimated dimensions for these areas are 100 m x Navacerrada transmitter was the only one used in 100 m. this campaign while the other one was switched off. The network architecture has been described in detail in [2] and [3].

The radiating system consists on four simple panels Analog propagation models assume lognormal with horizontal polarization and orientated 120 signal spatial variation. The standard deviation degrees NE. The tower height is 20 meters a.g.l. and parameter for this distribution has been considered the gain pattern was designed to ensure adequate to be from 9 to 15 dB [8]. Broadband signals (e.g. coverage in the northwest urban and suburban digital audio broadcasting) have been measured in areas of Madrid. The 26 UHF channel (center countries like UK and the shape of their statistical frequency 514 MHz) has been used to broadcast the variation has been proven to behave log-normally. DVB-T signal and the radiated power is 250 W. However, standard deviation for such signals has turned out to be significantly smaller [9]. The target DVB-T standards describe two different modes of of this measurement campaign is to obtain reference operation (8K and 2K). The VIDITER project was values for the equivalent field strength spatial focused on the 8K system, this one being the choice distribution standard deviation in digital terrestrial for future networks in Spain. Transmission television (COFDM-8K) networks. parameters like carrier modulation, Viterbi coding and guard interval are shown on table I. MEASUREMENT SYSTEM

The measurement campaign has been carried out System 8K using a specially designed mobile unit. This vehicle Number of carriers 6817 has a hydraulic mast able to lift an up to 10 Bandwidth 7.61 MHz m a.g.l. The antenna (Yagi-Uda type) has also been specially designed for the measurement campaign to Carrier Modulation 64 QAM receive DVB-T signals and its optimal frequency Inner code rate 2/3 range is centered on TV UHF channel 26 (510-518 Guard interval 1/4 MHz). Gain for this antenna is 11 dBi.

Table I. Transmitter operating mode. The following compose the measurement system:

It is envisaged that this particular COFDM mode ? RF Step attenuator will be the most suitable to implement future fully ? RF Switch operational networks in most European countries. ? ESVB field strength meter with a very sharp 8 This mode allows a useful throughput of about 20 MHz bandwidth filter MBPS and this bit rate is wide enough to broadcast ? Spectrum Analyzer four video and associated audio programs with a ? DVB-T Tuner/COFDM decoder. This decoder is subjective quality similar to the one offered by PAL a prototype developed by ITIS as one of the work systems. The UHF 8 MHz channel has also been aspects of a European digital television selected as representative of the frequency range to consortium. This receiver has been designed for be used all over Europe for DVB-T services [4]. this type of system validation measurement campaigns and allows access to any parameter MEASUREMENT CAMPAIGN TARGETS of the received signal [10]. ? B.E.R. meter DVB-T network planning has to be based on small ? GPS receiver area equivalent field strength spatial variation distribution studies. Dimensions for these small The data presented in this paper were obtained with areas have been established in 100 m x 100 m [5]. A an attenuation of 0 dB. This device and the DVB-T high percentage (99% for “good” and 70% for demodulating equipment were used to obtain C/N “acceptable” coverage) of the locations inside the ratios. Such data have been published in [11]. small area must achieve the required minimum C/N in order to ensure proper service quality. In order to be able to take measurements while the van was moving a special magnetic device attached As far as the received power is concerned, it could be to one of the vehicle wheels was used as a trigger stated that if the receiver figure is frequency signal source. All the data acquisition system was independent and high enough to neglect the input controlled by software running on a laptop allowing thermal noise over UHF band, the probability of easily setting up, customizing and storing all the receiving an equivalent field strength level higher parameters tested at each location. Figure 1 shows a than a certain amount must be over 99% (or 70%) block diagram of the measurement system installed [6][7]. in the vehicle.

device to send a trigger signal as the van moved Antenna Attenuator forward every 25 cm. This data acquisition technique allowed us to take measurements while GPS driving the vehicle at 10-20 Km/h along 100 m Switch distances towards the transmitter. The number of Field-Strength DVB-T Tuner measurements taken at each of these 100 m drives ESVB meter COFDM Receiver was approximately 1000. Figure 2 shows measurements taken in one of the 100 m drives. Spectrum Analyzer Horizontal axis shows data number and vertical one B.E.R. meter equivalent field strength values.

Measurement Personal Trigger signal Computer Control & Data adquisition

Figure 1. Measurement equipment.

MEASUREMENT TECHNIQUES

First step in this measurement campaign was to carefully choose the test points inside the expected coverage area of the Navacerrada transmitter. This choice was based upon several criteria. First, all the selected measurement points were to be located at first Fresnel ellipsoid obstruction free areas. Some Figure 2. Measurement drive example. computer-aided simulations were made using a digital terrain database to evaluate the Fresnel first There was an obvious risk in using such a ellipsoid obstruction for every location inside the measuring technique. Depending upon the location transmitter service area. characteristics, wherever ground irregularities exist, the variation between different measurements Expected median equivalent field strength has been in the same 100 m drive could have been caused by another requirement for considering data collecting changes in antenna orientation and mast places. Areas situated near the foreseeable coverage movement. Every 100 m drive was carefully limits were rejected in order to obtain reference observed to be able to reject the ones suspected to values for a general case planning. In consequence, have signal variations caused by the measurement spatial variation of the equivalent field strength method itself rather than by intrinsic electric field data inside small areas appeared to be influenced spatial changes. only by multipath effects at the receiver side, thus allowing the measured data to be considered as a RESULTS AND DISCUSSION statistically homogeneous ensemble. According to the explained criteria 36 different Finally, this preliminary study was restricted to zones were selected inside the expected coverage suburban environments, having scattered buildings area. Figure 3 shows the median equivalent field inside 100 m x 100 m areas. The receiving antenna strength for the set of measurements obtained at is roughly at the same height as the surrounding these locations. roofs. Residential areas all over Europe can be considered of this kind. Former studies carried out within the VIDITER and other experimental networks estimated the In order to achieve reliable equivalent field strength minimum equivalent field strength values required variation parameters it is necessary to take a high for achieving C/N ratios higher than the minimum number of measurements inside the 100 m x 100 m threshold for the COFDM mode described in this area. The technique to obtain this high recordset of article. These figures are near 50 dB?V/m and measurements used a special magnetic device correspond to a measured C/N of 21 dB [6][11]. installed on one of the wheels of the van to send trigger signals to the measuring equipment through Recordsets corresponding to the 100-m van drives, a parallel interface. It was possible to configure the where antenna movement or errors in antenna whole system to capture a data for different wheel orientation were present, were discarded. As a rotation angles. Data were collected adjusting this result of this filtering, the final data set was

composed of every measurement taken in 30 of the MEAN MEDIAN STANDARD HIGHEST AREA VALUE 36 locations. The mean equivalent field strength VALUE DEVIATION VARIATION ID measured inside each area was calculated, and the (dBuV/m) (dBuV/m) (dB) (dB) value obtained was used to calculate the difference 1 77,987 78,100 1,69 4,98 in decibels between the mean signal level and the 2 77,395 77,390 1,53 3,42 value of each point in the area. Table II shows the 3 76,796 78,300 3,37 13,79 standard deviation of the measured signal level 4 71,604 71,370 2,25 4,77 inside each area. The table also shows the mean and 6 67,108 67,210 2,13 6,02 median values as well as the maximum variation. 7 67,591 67,480 1,87 5,76 8 69,656 69,830 2,13 9,95 9 73,172 73,040 4,00 11,80 These received equivalent field strength variations 10 73,599 72,430 3,03 4,28 corroborate the need for small area based planning 12 84,046 83,930 1,68 4,07 when designing DVB-T networks. Otherwise, areas 15 77,606 78,140 2,55 7,96 where the median signal level is a few decibels 17 71,307 71,410 2,00 4,42 18 52,126 52,080 2,51 5,36 above the minimum that ensures the required C/N 19 81,257 81,320 0,62 2,11 ratio could have a high percentage of locations 20 80,909 81,080 1,06 3,25 where MPEG transport stream would be 21 76,526 76,800 1,64 4,12 unsuccessfully decoded. 22 78,003 77,950 1,18 2,45 23 61,300 61,300 1.29 3.53 24 81,058 80,920 0,85 1,7 Another analysis was made by considering all the 25 69,878 70,010 2,93 9,19 deviations from the mean level in each area as a 26 67,197 67,270 0,74 2,32 whole statistical ensemble. The restrictions applied 27 52,824 53,020 2,18 6,70 in the area selection process, along with the filtering 28 58,234 58,340 1,13 2,80 of collected data allowed us to do so. The deviations 29 49,036 48,680 1,18 1,74 30 59,991 59,690 2,43 4,63 made up a homogeneous statistical whole as far as 31 71,648 71,720 1,56 5,43 environment type and reception conditions, such as 33 68,859 69,170 2,42 9,13 first Fresnel ellipsoid obstruction and antenna 34 70,017 70,500 2,69 6,18 orientation, were concerned. 35 79,011 79,680 2,37 7,86 36 69,662 69,580 1,24 3,81

Table II. Median, mean value, standard deviation and maximum variation inside each area

Figure 3. Mean measured equivalent field strength inside each area.

Table III shows standard deviation and maximum Moreover, if the information is spread all over the value for the entire measurement campaign. signal bandwidth the errors caused by multipath fading affect only to a small part of the whole PARAMETER VALUE (dB) bitstream.

Standard Deviation 2.07 These two features make the performance at the receiver side appear very similar to the one of a Maximum variation 13.79 system with frequency diversity.

Therefore, bandwidth increase is directly related to Table III. Standard deviation and maximum the spatial distribution standard deviation decrease. variation for the whole set of measurements in the Variations for different frequencies behave more 30 finally selected areas. randomly as the distance between components is

increased, so the wider the spectrum is, the more Considering all the measurements made inside the effective the compensation. 100 m x 100 m areas it can be concluded that the equivalent field strength spatial variation Figure 4 shows the equivalent field strength spatial distribution approximates reasonably a log-normal variation distribution histogram. The horizontal distribution with a median value near to zero and a axis represents the deviation from the average value standard deviation of about 2 decibels. inside the small area and the vertical one the

number of measurements corresponding to that This significant decrease of the standard deviation variation interval. The bin is 0,2 decibels wide. The for small coverage areas is caused by the intrinsic obtained results are coherent when compared with nature of the multipath propagation. It is a well- measurements and studies made for other different known fact that multipath effects at the receiver broadband services such as digital audio side are frequency selective. If the signal being broadcasting. Specifically, measurements of the analyzed has a wide bandwidth, as it happens with DAB signal, whose bandwidth is four times smaller COFDM-8K signals, the phase delays of the than the DVB-T one, have shown that the standard different components of the signal show small deviation inside suburban environment areas is correlation. As the phase delays are not correlated near 4 dB [9]. the overall effect of multipath fading on the received field strength is smaller if compared with narrowband signals.

Distribution of Equivalent Electric Field spatial variation (dB)

2000

1800

1600

1400

1200

1000

800

600

400

Number of measurements 200

0 -9,0 -8,4 -7,8 -7,2 -6,6 -6,0 -5,4 -4,8 -4,2 -3,6 -3,0 -2,4 -1,8 -1,2 -0,6 0,0 0,6 1,2 1,8 2,4 3,0 3,6 4,2 4,8 5,4 6,0 6,6 7,2 7,8 8,4 9,0 Variation Interval

Figure 4. Equivalent field strength spatial variation distribution histogram.

CONCLUSIONS [2] P. Cañizares, J.L. Torres, J.A. Martínez “VIDITER: Spanish experience on DTT” The advanced state-of-the-art of digital terrestrial Proceedings of 20th International Television television network has been demonstrated by Symposium, Montreux , 13-17th June 1997. projects like VIDITER. These activities have included several network configuration tests and [3] P. Cañizares, J.L. Torres, S. Mata. “The first measurement campaigns. Spanish experience on digital terrestrial television broadcasting” IBC (International In this article a study of the spatial variation of the Broadcasting Convention), IEE No. 428, 1996. equivalent field strength has been presented for the COFDM 8K system. The COFDM operating mode [4] European Technical Standards Institute. T.R.101 parameters included 64 QAM carrier modulation, 190 “Implementation guidelines for DVB 2/3 Viterbi coding and a guard interval of 1/4. This systems; Transmission aspects”. December particular parameter combination is expected to be 1997. one of the best candidates for future broadcast networks that will be fully operative by the end of [5] AC106 VALIDATE Project Report. 1999 in some European countries. The results have “Implementation guidelines to DVB-T”. January corroborated former studies dedicated to systems 1997. with similar signal features (such as DAB) but belonging to different services. The most remarkable [6] C. Weck, R. Schramm.”Receiving DVB-T: Results conclusion is that field measurements on a real of field trials and coverage considerations”. DVB-T network have proved the expected decrease Montreux International TV Symposium. June of the standard deviation for the equivalent field 1997. strength variation distribution when compared with values. [7] CEPT Interim Report: ”Planning and introduction of digital terrestrial television In the same way the need for planning methods (DVB-T) in Europe. CEPT/FM-PT24, WGA 11 based on the statistical behaviour of the DVB-T Rev 2, WGB 14, May 1996. signal inside small 100 m x 100 m area has been also assessed. Although the decrease of the signal [8] ITU-Recommendation PN.370-5: VHF and UHF level variation compared with the analog case could propagation curves for the frequency range from inspire a less restrictive planning, the narrow 30 to 1000 MHz; Broadcasting Services. decibel margin from an error free decoding to an unsuccessful MPEG-TS recovery makes the [9] M.D.C. Maddocks, I.R. Pullen, J. A. Green, “Field traditional planning methods unsuitable. trials with a high-power VHF single frequency network for DAB”. EBU Technical Review. Finally it can be concluded that although some Autumn 1994 preliminary reference values have been obtained more extensive measurement campaigns should be [10] AC106-VALIDATE Project Report: CCETT & carried out to set up the final minimum equivalent ITIS. “DVB-T Multidimensional Tests: Test field strength values that ensure correct C/N ratios Results”. June 1998 and standard deviations for spatial variation in small areas. These field studies based data will be [11] A. Arrinda, M Vélez, P. Angueira., D. De la critically important for correct future planning. Vega, J.L. Ordiales. “Digital terrestrial television (COFDM-8K system) field tests and ACKNOWLEDGEMENTS coverage measurements in Spain”. IEEE Trans. On Broadcasting. Vol. 45. Nº 2 June 1999. We would like to thank all the participants in the VIDITER project and specially RETEVISION, which has made all this work possible.

REFERENCES [1] ETSI (European Telecommunications Standards Institute) “Digital Video Broadcasting (DVB); Framing structure, channel coding and modulation for digital Terrestrial television (DVB-T)” ETS 300 744 March 1997.