IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 3, SEPTEMBER /$ IEEE

Transcription

1 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 3, SEPTEMBER Procedure to Optimize Coverage and Throughput for a DVB-H System Based on Field Trials Wout Joseph, Member, IEEE, David Plets, Leen Verloock, Emmeric Tanghe, Luc Martens, Member, IEEE, Etienne Deventer, and Hugo Gauderis Abstract A procedure to optimize the coverage and throughput for a DVB-H system is proposed. The performance of the system is evaluated by measurements for different modulation schemes, guard intervals, and MPE-FEC rates. Based on technical trial results, an optimal transmission scheme is proposed for a specific network, maximizing the coverage for a certain throughput requirement. This optimization will enable to select an optimal transmission scheme for future DVB-H trials and deployments. Index Terms Coverage, digital video broadcasting-handheld, DVB, DVB-H, indoor reception, mobile reception, MPE-FEC, optimization, throughput. I. INTRODUCTION T HE DIGITAL broadcasting standard DVB-H (Digital Video Broadcasting-Handheld) enables a high data rate broadcast access for hand-held terminals (e.g., portable, pocket-size battery-operated phones) [1] [4]. The broadband downstream channel features high data rates and may be used for audio and video streaming applications, file downloads, and many other kinds of services. The DVB-H technology is an extension of DVB-T (Digital Video Broadcasting-Terrestrial) and takes the specific properties of typical hand-held terminals into account. The three main new physical-layer techniques that have been introduced for DVB-H are time-slicing, MPE-FEC, and the 4K mode [1], [2]. First, DVB-H uses time slicing, a power-saving algorithm based on the time-multiplexed transmission of different services. This technique results in a battery power-saving effect and allows soft handover if one moves from a network cell to another one. Secondly, for reliable transmission in poor signal reception conditions, an enhanced error-protection scheme, called MPE-FEC (Multi-Protocol Encapsulation Forward Error Correction), is introduced. Thirdly, the 4K mode (next to the 2K and 8K mode of DVB-T) for OFDM (orthogonal frequency division multiplexing) is defined, addressing the specific needs of hand-held terminals. The 4K mode aims to offer an additional trade-off between transmission cell size and mobile reception capabilities, providing an additional degree of flexibility for DVB-H network planning for single-frequency networks (SFN). These techniques make DVB-H a very promising standard for broadcast services requiring high data rates and offers extended possibilities for content providers and network operators. The purpose of this paper is to select an optimal set of parameters to obtain typical and necessary data rates for mobile TV-services. The optimization procedure will be demonstrated for bit rates of 5 and 10 Mbps for inside car and indoor building reception. Up to now transmission schemes are selected in a more arbitrary way, taking into account bit rate requirements. To obtain the optimal scheme, the performance of a DVB-H system is evaluated using actual Flemish trial results [5] [7]. Also results of other trials in other countries will be considered [3], [8], [9]. Tests are performed in a rural and suburban environment for different reception conditions. Mobile reception in car at different speeds and indoor portable reception are here considered. The influence of the settings on the (carrier-to-interference-noise ratio) will be determined. The range of the system will then be optimized by varying the different system settings (modulation, guard interval, MPE-FEC coding rate), and by analyzing the resulting performance and ranges for the different DVB-H characteristics. The methodology and results of this paper can be used for the selection of the optimal transmission scheme in future field trials and commercial deployments. The outline of this paper is as follows. The configuration and different settings of the DVB-H system for the optimization are described in Section II. In Section III, the methodology, investigated scenarios, and analyzed quality parameters are discussed. Optimization of the parameters of the DVB-H system by investigation of coverage and link budget is performed in Section IV. Finally, the conclusions are presented in Section V. Manuscript received November 8, 2007; revised March 27, First published June 10, 2008; last published August 20, 2008 (projected). This work was supported by the IBBT MADUF project, cofunded by the IBBT (Interdisciplinary institute for BroadBand Technology), a research institute founded by the Flemish Government in 2004, and the involved companies and institutions. W. Joseph is a Post-Doctoral Fellow of the FWO-V (Research Foundation-Flanders). W. Joseph, D. Plets, L. Verloock, E. Tanghe, and L. Martens are with the Department of Information Technology, Ghent University/IBBT, Gent 9050, Belgium ( wout.joseph@intec.ugent.be). E. Deventer and H. Gauderis are with the Flemish Radio and Television Network (VRT), Brussel 1043, Belgium. Color versions of one or more of the figures in this paper are available online at Digital Object Identifier /TBC A. Transmission System II. CONFIGURATION Fig. 1 shows a map of Ghent where the Flemish DVB-H trial network is situated and the location of the three base stations (BS) marked with large black dots. Other labels and colors will be explained further. The environment is rural and suburban. The terrain is flat. The operating frequency is 602 MHz with a bandwidth of 8 MHz. In Belgium, there are almost no urban regions. Urban is defined in this paper as an area as the center of New York, only parts of the town Brussels in Belgium are /$ IEEE

2 348 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 3, SEPTEMBER 2008 TABLE I TRANSMISSION SCHEMES AND CORRESPONDING PHYSICAL BIT RATE, INVESTIGATED TO DETERMINE THE INFLUENCE OF MPE-FEC, MODULATION SCHEME (AND INNER CODE RATE), AND THE GUARD INTERVAL Fig. 1. Map of Ghent with the three transmitting antennas (large black dots), the selected buildings for the indoor portable measurements (blue dots), and the routes for the mobile measurements (the routes at 20 km/h are shown in red, the routes at 70 km/h are shown in green). urban. Cities as Ghent can be classified as suburban, the region around the city center can be classified as rural. All transmitting antennas are omnidirectional and are vertically polarized. The heights of these BS are (BS3), and (BS1), (BS2), respectively. The EIRP used for these BS is 39.0 dbw, 36.6 dbw, and 40.9 dbw, respectively. Because the 4K mode is exclusively defined for use in DVB-H systems we investigate here the performance of this 4K mode. Time synchronization is achieved by Meinberg GPS receivers with a 10-MHz clock. The absolute accuracy is 1. The 10-MHz clock is also used to synchronize the transmitting frequency of the different transmitters in the SFN. In the network no static delay is used i.e., all transmitters transmit at the same time. B. Investigated Transmission Schemes for Optimization The parameters that have been tuned are modulation, coding rate, guard interval GI (ratio of guard time and useful time ), and MPE-FEC coding rate. For each set or combination of parameters we define a number (no). A transmission scheme will be noted as follows: FFT mode, guard interval, modulation and inner code rate, MPE-FEC rate. Table I gives an overview of the 14 investigated transmission schemes and their corresponding physical (PHY) bit rate. The parameters that are varied are shown in bold in Table I. Six parameters sets (no 1-6) have been selected to investigate MPE-FEC rates ranging from 67/68 to 1/2. For six parameter sets, different modulation schemes and inner coding rates from QPSK 1/2 to 64-QAM 2/3 are selected (no 7-11 and no 2). Finally, to study the influence of the guard interval (1/32 up to 1/4) four schemes are selected as shown in Table I. III. METHOD A. Investigated Scenarios 1) Portable Reception: Indoor walking has been investigated. Measurements have been executed during one minute on each floor in every building. In total 13 buildings have been investigated in different environments. In Fig. 1 the different buildings are marked with blue dots. Only 11 locations are shown in Fig. 1: at two locations, two buildings are investigated, resulting in a total of 13 buildings. These buildings are located in different environments: in dense suburban areas (area with narrow streets (width less than 10 m) and terraced houses), in suburban areas (the streets are wider, with a width between 10 and 15 m), in residential environments (width of the streets is more than 15 m and most houses have gardens), and in rural environments (mostly open, at least on one side of the street, and contains only few houses). 2) Mobile Reception: For mobile reception (class D reception inside a vehicle [4]), measurements have been performed for reception inside a car at different reception velocities: 20 km/h and km/h (maximum allowed speed in Ghent). Six routes at 20 km/h have been chosen: one in each wind direction, one through the city center and one around the city. The total length of the routes at 20 km/h is 70.5 km. The total length of the routes at km/h is 37.5 km. Fig. 1 shows the routes for investigating the different mobile scenarios of class D. The routes at 20 km/h are shown in red, the routes at 70 km/h are shown in green. B. Parameters Used for Analysis An important criterion for good reception is based on the MFER or Multi-Protocol Encapsulation Frame Error Rate [3]. MFER is the ratio of the number of residual erroneous frames (i.e., not recoverable) and the number of received frames [3]: MFER (1) Results from [9] showed that the MFER 5% objective criteria corresponded to a good/fair recovery of audiovisual programs, subjectively reported by two observers [3]. It has

3 JOSEPH et al.: OPTIMIZING COVERAGE AND THROUGHPUT FOR DVB-H SYSTEM BASED ON FIELD TRIALS 349 also been shown that an MFER 10 % corresponds to annoying recovery. Furthermore, FER (Frame Error Rate) is the ratio of the number of erroneous frames before MPE-FEC correction and the number of received frames [3]. In this paper, we will use as signal quality requirements: : the minimal value of for which the MFER is at most X%, or %Locations (i.e., the ratio of the number of valid frames received in a 1 db range for and the total number of received frames [3], [7]) is at least 100-X after MPE-FEC correction (X will be 5). : the minimal value of for which the FER is at most X%, or %Locations is at least 100-X before MPE-FEC correction (X will be 5). The MPE-FEC gain, noted as, is then the reduction in db for obtained by using MPE-FEC, while maintaining the same reception quality (X will be 5): C. Measurement Method The measurements are performed with a DVB-H tool implemented on a PCMCIA card with a small receiver antenna [7]. The antenna is a Pulse DVB-H MHz Planar PWB (planar printed wire board) antenna with the following dimensions: length of 50.5 mm, width of 10.5 mm, thickness of 3.0 mm. The gain of the system is 5 dbi. The connector is of type MMCX. The PCMCIA card is plugged into a laptop, which is used to collect and process the measurements later. Every 0.5 s, a sample is recorded, while the receiver is either locked or unlocked [5] [7]. A locked receiver can receive MPE-FEC frames, which are either correct or incorrect. Incorrect frames can be corrected by the MPE-FEC code. The tool logs parameters as (carrier-to-interference-noise ratio), FER (Frame Error Rate), MFER (Multi-Protocol Encapsulation FER), and electric-field strength. Location and speed are recorded with a GPS device. During the measurements, the video channel één of the public broadcaster VRT (Flemish Radio and Television Network) is monitored. The investigated modulation schemes (see Section II-B) are broadcast with 768 and 512 rows. Using the right packet identifier, the receiver can stream a channel of the transmitted DVB-H signal. By opening a session description protocol (sdp) file, we can monitor the channel on the laptop with a media player. For mobile reception, measurements are performed inside a small van, driving around at different velocities. D. Path Loss Models To calculate the range of the system (and select the optimal parameter setting), path loss models are needed. The path loss (PL) between a pair of antennas is the ratio of the transmitted power to the received power. It includes all of the possible elements of loss associated with interaction between the propagating wave and any objects between the transmitting and receiving antennas [10]. The parameter PL is used for the estimation of the coverage of a system. Different models to obtain the (2) Fig. 2. Path loss at 602 MHz as function of the distance from the BS for ITU and Ghent models (suburban environment, h =60mand h =2:85 m). path loss have been developed. We discuss here two models that are applicable for the estimation of the coverage of DVB-H systems: the ITU-R P.1546 model [11] and our own Ghent model [5], [12]. 1) ITU-R P.1546 Model: The ITU-R P.1546 model [11] uses tabulated field-strength values for a 1 kw equivalent radiated power at nominal frequencies of 100, 600, and 2000 MHz. From the field values the path loss can be calculated. The described propagation curves represent the field-strength values exceeded for 50 %, 10 % and 1 % of time. The ITU-R P.1546 model [11], noted as ITU model in this paper, is not valid for field strengths exceeded for percentage times t outside the range from 1 % to 50 %. The ITU model will be used for rural and suburban environments in this paper. The range of the models is larger than 1 km. The ITU-R P.1546 model gives a standard deviation of 5.5 db for wideband signals [11]. 2) Ghent Model: In [5] and [12], the path loss is determined using measurements of an actual DVB-H signal at 602 MHz in a suburban environment. This model will be noted here as the Ghent model. The path loss is modeled according to a lognormal shadowing model: where d is the distance between BS and Rx in m, is a reference distance in m, and n is the path-loss exponent. For, 100 m was chosen here. Furthermore, is the shadowing fading variation and has a standard deviation.afit with two parameters, and n, was performed. The root-mean-square (RMS) deviation of the measurement points was minimized with a linear regression fit. The parameter equals 86.8 db. We obtained a path-loss exponent and a standard deviation of 6.18 db. It was shown in [12] that the variation around the mean path loss is well described by a lognormal distribution. This model is valid from 70 m up to 14 km, and for BS heights around 60 m. Fig. 2 shows the path loss at 602 MHz for the ITU and Ghent models for a suburban environment. The height of the BS is 60 m and the height of the receiving antenna is 2.85 m. The height correction of [11] can then be used to obtain values at. Up to 1.96 km, the Ghent model delivers the (3)

4 350 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 3, SEPTEMBER 2008 TABLE II SHADOWING MARGINS USED IN THIS PAPER BASED ON ETSI TR [4] highest path losses, resulting in the most restrictive model. From that distance on, the ITU model is more restrictive (the environment around Ghent is then less dense suburban). 3) Shadowing Margins: In this paper all ranges will be calculated for good reception. This good reception is defined as follows: at least 95% of receiving locations at the edge of the area are covered for portable reception and 99% of receiving locations within it are covered for mobile reception [4]. The standard deviation of the ITU model is used in [4] for determining the shadowing margin (noted as location correction factor in [4]) for outdoor locations (portable and mobile). The shadowing margin at indoor locations is the combined result of the outdoor variation and the variation factor due to building attenuation. These distributions are expected to be uncorrelated. The standard deviation of the indoor field strength distribution can therefore be calculated by taking the root of the sum of the squares of the individual standard deviations. At UHF, the outdoor and indoor macro-scale standard deviations are 5.5 db and 6 db according to [4], [11], respectively. The combined value is then 8.3 db. Table II shows the shadowing margins used in this paper and based on ETSI [4]. We assume here a building penetration loss of 11 db as proposed in [4]. We assume a vehicle entry loss of 8 db and a standard deviation of 2.5 db [13]. The combined value of the outdoor and vehicle entry deviation is then 6 db. The shadowing margin for mobile reception in a car (class D) is then 14 db for 99 % coverage. IV. OPTIMIZATION In this section, the range of the DVB-H system is optimized by changing the different parameters of the transmission schemes. By varying the guard interval, the MPE-FEC coding rate, and modulation scheme and inner code rate, we will be able to optimize the range of the DVB-H system for a certain data rate. We demand here a throughput of about 5 and 10 Mbps (corresponding with about 27 video channels for 10 Mbps). Fig. 3 shows a flow graph of the procedure to optimize the range of the system. First, the different parameters are varied and the and MPE-FEC gains are determined for 14 different transmission schemes, taking into account the MFER 5% criterion of [4] discussed in Section III-B. These results will be presented in Sections IV-A IV-C. Next, the link budget for different transmission schemes is calculated using both the path loss models of Section III-D and the results of the field tests (MPE-FEC gain,, etc.). Then the range of the different schemes is calculated and finally, the requirement of 5 and 10 Mbps is taken into account to select the optimal transmission scheme with a maximal range. The different steps will now be discussed. Fig. 3. Flow graph of the optimization procedure. A. MPE-FEC The MPE-FEC interleaves the data and adds a Reed-Solomon Code in order to make the DVB-H signal more robust. The influence of the MPE-FEC coding rate is also investigated in [3], [8]. In this paper we use the schemes of Table I (Section II-B) to analyze experimentally the influence of the MPE-FEC. Investigated coding rate levels are 1/2, 2/3, 3/4, 5/6, 7/8 and 67/68. A decreasing level of protection corresponds to a higher coding rate level. The coding rate level 67/68 corresponds to almost no MPE-FEC protection. In this section the parameters 4K, 1/8, 16-QAM 1/2 are fixed and the MPE-FEC rate is varied. We will assume further that the MPE-FEC gains are about the same for all considered modulation schemes (e.g., differences smaller than 1 db for MPE-FEC gain when changing from QPSK to 16-QAM [6], [7]). Table III shows the value of the MPE-FEC gain for the different MPE-FEC rates for reception inside car at 20 km/h and 70 km/h and indoor walking reception, determined using the method of Section III [6], [7]. As expected, Table III shows that the more MPE-FEC coding is used (lower MPE-FEC rate), the higher the gain in [db]. The MPE-FEC rate 7/8 provides a somewhat larger gain than expected but corresponds reasonably with data of [14] where gains from 1.6 db (at Doppler frequency of 10 Hz, i.e., portable) up to 4.3 db (at about 250 Hz) are obtained for a typical urban (TU6) channel. Our values are based on actual measurements (realistic channel) inside a car. For 67/68, the gain is very limited and maximally about 0.57 db. The highest gain for MPE-FEC 7/8 is obtained for scenario indoor walking (2.62 db). For 5/6, the highest gain is 1.36 db for car 70 km/h reception. These gains are lower than for 7/8. For MPE-FEC 3/4 gains of 1.94 and 2.96 db are obtained for car 20 and 70 km/h. Finally, for 1/2 maximal gains of about 2.70 db are obtained. In [15], an MPE-FEC gain (constellation 16-QAM 1/2, MPE-FEC rate 3/4) of 1.5 db for a car driving at 30 km/h has been obtained, when measuring at an IP PER (Packet Error Rate) of 5 %. This corresponds reasonably well to the

5 JOSEPH et al.: OPTIMIZING COVERAGE AND THROUGHPUT FOR DVB-H SYSTEM BASED ON FIELD TRIALS 351 TABLE III 4C VALUES FOR NINE SCENARIOS AND DIFFERENT MPE-FEC RATES FOR DIFFERENT SCENARIOS (4K, 1/8, MODULATION SCHEME FIXED AT 16-QAM 1/2) MPE-FEC gains obtained in Table III for an MPE-FEC rate of 3/4 (a gain of 1.94 db for a car driving at 20 km/h). In [8] it is stated that already in portable situations (Doppler frequency below 10 Hz), the effect of the MPE-FEC rate allows to reach improved. The effect of the MPE-FEC gradually improves the for higher Doppler frequencies e.g., for code rates 3/4 and 5/6 in Table III. B. Modulation Scheme In Section II-B (Table I) the investigated transmission schemes are shown. Fig. 4 shows the and values for the investigated modulation schemes for indoor walking reception and for reception inside a car at 20 km/h [7]. These figures show that the higher the modulation scheme, the higher the required values: e.g., in Fig. 4(b) for scenario IV (car 20 km/h) for QPSK 1/2, a value of 7.28 db is required, while for 64-QAM 2/3, db is required. The required MFER 5% values (black bars in Fig. 4) are lower than the FER 5% values (white bars in Fig. 4), due to the MPE-FEC coding gain. In Fig. 4(b), the FER 5% value is missing for 64-QAM 2/3: reception without MPE-FEC coding is not possible with the receiver under test. The required value is higher than the value that can be measured with the used receiver. Table IV (Fig. 3) shows the and values for all scenarios and for all modulation schemes. It shows again that the higher the modulation scheme, the higher the required value. When the inner coding changes from 1/2 to 2/3, the value is 2 to 3 db higher. When the constellation is 16-QAM instead of QPSK, the value is about 6 db higher. When changing from 16-QAM to 64-QAM, an increase in of about 4 db is needed for indoor walking to maintain the same reception quality. In [4], [8], required values of 15.5 db, 18.5 db, 20.5 db, and 23.4 db for 16-QAM 1/2, 16-QAM 2/3, 64-QAM 1/2, and 64-QAM 2/3 respectively, are presented for mobile reception. These values correspond reasonably well with our measurements for the scenario of car 70 km/h (see Table IV), where the obtained values are db, db, db, and an unmeasurable value (the required value for 64-QAM 2/3 is higher than the value that can be measured with the used receiver). For 64-QAM 2/3 we use as value for mobile reception at 70 km/h 23.4 db [8], [14]. For QPSK 1/2 and QPSK 2/3 for car 70 km/h also values of [8], [14] are used. The differences between the values from our measurements and those of [8], [14] may be caused by the lower Doppler shift in our measurements, compared to the mobile (higher velocity) situation in [4], [8], [14]. It Fig. 4. C=(N + I)j and C=(N + I)j values for different modulation schemes (a) for indoor walking reception and (b) for reception inside a car at 20 km/h (4K, 1/8, MPE-FEC fixed at 7/8). should also be noted, that in [4], [8], the MPE-FEC rate is probably 3/4 and the channel is typical urban (TU6 channel), while in this paper, the MPE-FEC rate is 7/8 for an actual (realistic) channel (inside a vehicle). The MFER 5% values of Table IV will be combined with those of Table III to optimize the range of the system (see also Fig. 3). C. Guard Interval In order to avoid inter-symbol interference (ISI) and inter-carrier interference (ICI), a guard interval GI is introduced in front of every data part of an OFDM symbol of the system. By decreasing the guard interval one can increase the physical bit rate but one has to take into account that the lower guard interval offers less protection for ISI and ICI (delay spread). Also the maximum acceptable echo delay depends on the used guard interval. When the echo delay of the signal is higher than the guard interval then interference occurs. The maximum distance

6 352 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 3, SEPTEMBER 2008 TABLE IV THE REQUIRED (C=N +I)j FOR THE CONSIDERED MODULATION SCHEMES FOR DIFFERENT SCENARIOS (4K, 1/8, MPE-FEC FIXED AT 7/8) TABLE V C=(N + I)j AND C=(N + I)j VALUES FOR DIFFERENT GUARD INTERVALS FOR CAR 20 km/h (4K, 16-QAM 1/2, 7/8) among the transmitters in an SFN, which would produce minimal inner interference, is defined as follows [4]: Where c is the velocity of light and is the guard time. In practice, multiple path signals will be received simultaneously coming from one transmitter with different delays and amplitudes and interference is inevitable. Scenario car 20 km/h has been investigated to study the influence of the guard interval. Table V shows the and maximal deviation for the different guard intervals. Table V shows that these values remain relatively constant for the different guard intervals, indicating that the influence of the guard interval on these values is small (maximally about 1 db). In Section IV-D the influence of the guard time on the will be neglected (method of Fig. 3). For the 4K mode, equals 4.2, 8.4, 16.8, and 33.6 km for guard intervals of 1/32, 1/16, 1/8, and 1/4, respectively [4], [9]. A guard interval of 1/32 is not considered in this paper because this setting can only be used for small SFNs ( equals for the 4K mode only 4.2 km), resulting in possible interference for the network under consideration (see Section II-A). D. Application of Optimization of Range In this section the optimal transmission scheme with maximal range will be determined for indoor reception (class B) and mobile reception (20 km/h, 70 km/h) for a realistic transmitter with an EIRP of dbw (about 8.2 kw),, and 602 MHz. The method (Fig. 3) can of course be used for any network with other EIRPs and heights, but this application is of interest of the Flemish DVB-H trial and will therefore be discussed here. The optimization of the range is performed by varying the guard interval, the MPE-FEC coding rate (67/68 up to 1/2), and the modulation scheme and inner code rate (4) (QPSK 1/2 up to 64-QAM 2/3), retaining a bit rate of about 5 and 10 Mbps (see also Fig. 3). We use the data from Tables III and IV and Fig. 4. We accept bit rates deviating maximally 10 %: and Mbps are thus the bit rate requirements, respectively. Without these requirements, 144 transmission schemes are possible i.e.,. Table VI shows alternative solutions for the and requirement and the corresponding ranges and bit rates. The optimal solutions are shown in bold. The results for the are only for small networks and are not shown here. For 10 Mbps, 16-QAM and 64-QAM modulations satisfy this requirement (Table VI). The highest ranges are obtained for 16-QAM 1/2, MPE-FEC 7/8: the maximal range for indoor reception (buildings) is about 2.2 km (ITU model) and 2.7 km (Ghent model) for both and 1/8 (in Section IV-C we have shown that the GI does not have a significant influence on the ). For mobile reception, the maximal ranges are higher and at e.g., 20 km/h about 2.6 (ITU) and 3.4 km (Ghent). The ITU and Ghent model correspond reasonably well. The Ghent model delivers mostly somewhat higher ranges. The Ghent model is less restrictive for ranges larger than 2 km, as mentioned in Section III-D, due to the more rural environment around Ghent. When the range is below 2.0 km, both models deliver more similar results. Our measurements show that indoor building reception is more difficult than reception into a car shown by smaller ranges for indoor reception. This is due to the higher building penetration losses than the vehicle penetration loss (about 4 to 8 db, [13], [16]). In [4] a building penetration loss of 11 db is assumed but often much higher penetration losses are recorded [17], [18]. Moreover, taking into account the considered practical speeds in Belgium and most of Europe (the maximum speed limit in Belgium is 120 km/h), no additional 3 db margin, as proposed in [4] and [19] is required for mobile reception in a car. The ranges for mobile car reception at 20 km/h and 70 km/h are often about the same: the required at 70 km/h is higher than at 20 km/h (see Table IV) but the MPE-FEC gain for 70 km/h is mostly higher than for 20 km/h (see Table IV), compensating sometimes the higher requirement. This results in about similar ranges for both mobile scenarios. It can be clearly seen in Table VI that the 64-QAM modulations also satisfy the 10 Mbps requirement, but the MPE-FEC gain cannot compensate the higher required. This results in lower ranges for indoor reception compared to the 16-QAM modulations: only 1.6 km for

7 JOSEPH et al.: OPTIMIZING COVERAGE AND THROUGHPUT FOR DVB-H SYSTEM BASED ON FIELD TRIALS 353 TABLE VI OPTIMIZATION FOR BIT RATES OF 10 Mbps AND 5 Mbps AND THE RESULTING RANGES FOR THE SCENARIOS UNDER CONSIDERATION 64-QAM 1/2 and 1.2 km for 64-QAM 2/3 are possible using the ITU model (see Table VI for ). If one has a network where can be lower than 8.4 km then the GI of 1/16 can be used but one has to take into account that the lower guard interval offers less protection for ISI and ICI (delay spread). The optimal scheme of Table VI for Mbps with maximal range is then:, 16-QAM 1/2, MPE-FEC 7/8, corresponding with Mbps, which is higher than for and therefore preferred. If a GI of 1/8 is necessary, the optimal scheme 16-QAM 1/2, MPE-FEC 7/8 corresponds with 9.68 Mbps. The ranges for the modulations for a are lower and maximally 1.9 (ITU) and 2.2 km (Ghent). For 5 Mbps, the ranges are of course higher than for the schemes at 10 Mbps. Lower modulation schemes in comparison to the ones for 10 Mbps satisfy the requirement: QPSK 1/2, QPSK 2/3, and 16-QAM 1/2 can be used (see Table VI). The scheme with a maximal range of 3.2 km (ITU) and 4.4 km (Ghent) for indoor reception (buildings) is QPSK 1/2, MPE-FEC 7/8 for and 1/8. These schemes are shown in bold in Table VI. For, somewhat lower ranges are obtained with maximal ones for QPSK 1/2, MPE-FEC 67/68. The optimal scheme for 5 Mbps for indoor reception is, QPSK 1/2, MPE-FEC 7/8 because this corresponds with the highest throughput of 5.12 Mbps. For and 1/8, for mobile reception (20 and 70 km/h) also the scheme QPSK 1/2, MPE-FEC 7/8 delivers the maximal ranges: for car 20 km/h ranges of 3.9 km (ITU) and 6.0 km (Ghent) are obtained, while for car 70 km/h ranges of 3.7 km (ITU) and 5.5 km (Ghent) are obtained. The optimal scheme for mobile reception is then also, QPSK 1/2, MPE-FEC 7/8 (5.12 Mbps) but also the scheme, QPSK 1/2, MPE-FEC 5/6 (4.87 Mbps) performs almost equally well for car 20 km/h and 70 km/h (about same ranges). Similar conclusions concerning the models and the different scenarios as for 10 Mbps can be drawn. We can conclude that by implementing and executing this procedure, the optimal schemes for a certain bit rate requirement can be selected. This procedure enables to select optimal schemes for future DVB-H trials and deployments. V. CONCLUSIONS In this paper a procedure to optimize the coverage and throughput for a DVB-H system is developed. The performance of a DVB-H system is first evaluated for different modulation schemes, guard intervals, and MPE-FEC rates by measurements. Indoor portable reception and mobile reception in a car is investigated. Based on technical trial results, an optimal transmission scheme then is proposed for a specific network, maximizing the coverage for a throughput requirement of 5 and 10 Mbps. For a requirement of 10 Mbps, 16-QAM 1/2 with MPE-FEC 7/8 delivers the largest ranges of about 2.2 km for indoor reception and 2.6 km for mobile reception for suburban environments (ITU model). For 5 Mbps, ranges of 3.2 km (indoor) and 4.0 km (mobile) are possible for QPSK 1/2, MPE-FEC 7/8. The influence of the guard interval on the range is limited. The methodology can be used for any DVB-H system. The results and analysis of this paper can be used for the selection of the optimal transmission scheme in future field trials and commercial deployments. REFERENCES [1] Digital Video Broadcasting (DVB); Transmission System for Handheld Terminals (DVB-H), EN v1.1.1, ETSI, Oct

8 354 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 3, SEPTEMBER 2008 [2] Digital Video Broadcasting (DVB); Framing Structure, Channel Coding and Modulation for Digital Terrestrial Television, EN v1.5.1, ETSI, Nov [3] Digital Video Broadcasting (DVB); Transmission to Handheld Terminals (DVB-H); Validation Task Force Report, TR v1.1.1, ETSI, May [4] Digital Video Broadcasting (DVB); DVB-H Implementation Guidelines, TR v1.1.1, ETSI, Feb [5] D. Plets, W. Joseph, L. Martens, E. Deventer, and H. Gauderis, Evaluation and validation of the performance of a DVB-H network, in 2007 IEEE International Symposium on Broadband Multimedia Systems and Broadcasting, Orlando, FL, USA, Mar. 2007, available on CDROM. [6] D. Plets, L. Verloock, E. Tanghe, W. Joseph, L. Martens, E. Deventer, and H. Gauderis, Evaluation of performance characteristics of a DVBH network for different reception conditions, presented at the 57th Annual IEEE Broadcast Technology Society Symposium, Washington, DC, USA, 31 October 2 November 2007, paper no , unpublished. [7] D. Plets, W. Joseph, L. Verloock, E. Tanghe, L. Martens, E. Deventer, and H. Gauderis, Influence of reception condition, MPE-FEC rate and modulation scheme on performance of DVB-H networks, IEEE Trans. Broadcasting (Special Issue on Quality Issues in Multimedia Broadcasting), accepted for publication. [8] G. Faria, j. A. Hendriksson, E. Stare, and P. Talmola, DVB-H: Digital broadcast services to handheld devices, Proceedings of the IEEE, vol. 94, no. 1, pp , Jan [9] T. Owens, C. Zhang, T. Itagaki, J. Outters, J. Lauterjung, M. Martucci, D. Bouquet, B. Mazieres, J. Prudent, P. Christ, I. Gaspard, S. Ritscher, G. Zimmermann, and P. Christ, Deliverable 6.1: Radio spectrum, Traffic Engineering and Resource Management Sept [Online]. Available: Tech. Rep. [10] S. R. Saunders, Antennas and Propagation for Wireless Communication Systems. New York, NY: John Wiley & Sons Inc., [11] Method for Point-to-Area Predictions for Terrestrial Services in the Frequency Range 30 MHz to 3000 MHz, ITU-R Recommendation P.1546, [12] D. Plets, W. Joseph, E. Tanghe, L. Verloock, and L. Martens, Analysis of propagation of actual DVB-H signal in a suburban environment, presented at the 2007 IEEE International Symposium on Antennas and Propagation, Honolulu, Hawaii, USA, June 2007, paper No. 1386, unpublished. [13] I. Kostanic, C. Hall, and J. McCarthy, Measurements of the vehicle penetration loss characteristics at 800 MHz, in Proc. 48th IEEE Int. VT Symp., May 1998, pp [14] TeamCast, DVB-H Calculator, [Online]. Available: teamcast.com/en/maj-e/c2a2i9177/support/dvb-h-calculator Team- Cast, DVB-H Calculator [Online]. Available: [15] H. Himmanen and T. Jekola, DVB-H field trials: Studying radio channel characteristics, in 2007 IEEE International Symposium on Broadband Multimedia Systems and Broadcasting, Orlando, Florida, USA, March [16] E. Tanghe, W. Joseph, L. Verloock, and L. Martens, Evaluation of vehicle penetration loss at wireless communication frequencies, IEEE Trans. Veh. Techn., May 2008, accepted for publication. [17] A. Turkmani and A. de Toledo, Modelling of radio transmissions into, and within buildings at 900, 1800 and 2300 MHz, IEE Proceedings, vol. 140, no. 6, pp , Dec [18] W. Joseph, E. Tanghe, D. Pareit, and L. Martens, Building penetration measurements for indoor coverage prediction of DVB-H systems, presented at the 2007 IEEE International Symposium on Antennas and Propagation, Honolulu, Hawaii, USA, June 2007, paper No. 1282, unpublished. [19] R. Schramm, DVB-T C/N values for portable single and diversity reception, EBU Technical Review, Wout Joseph (M 05) was born in Ostend, Belgium on October 21, He received the M. Sc. degree in electrical engineering from Ghent University (Belgium) in July From September 2000 to March 2005, he was a research assistant at the Department of Information Technology (INTEC) of the same university. During this period, his scientific work was focused on electromagnetic exposure assessment. His research work dealt with measuring and modeling of electromagnetic fields around base stations for mobile communications related to the health effects of the exposure to electromagnetic radiation. This work led to a Ph.D. degree in March Since April 2005, he is postdoctoral researcher for IBBT-Ugent/INTEC (Interdisciplinary institute for BroadBand Technology). Since October 2007, he is a Post-Doctoral Fellow of the FWO-V (Research Foundation-Flanders). His professional interests are electromagnetic field exposure assessment, propagation for wireless communication systems, antennas and calibration. Furthermore, he specializes in wireless performance analysis and Quality of Experience. David Plets was born in 1983 in Torhout, Belgium on the 26th of May. After an education in mathematics and sciences in secondary school, he began his engineering study in Ghent. Five years later, he finished his final year dissertation on the development of an idtv framework for sport coverage on the Multimedia Home Platform (MHP) and he obtained a Master in Electrotechnical Engineering, with ICT as main subject. Currently, he is a member of the WiCa research group (Department of Information Technology-INTEC, Ghent University), where he mainly focuses on DVB-H. Leen Verloock was born in Eeklo, Belgium on November 15, She received the degree of Master in electronics engineering from Katholieke Hogeschool Ghent (Belgium) in She then joined the Department of Information Technology (INTEC) of Ghent University where she is currently working as technical and research assistant in the Wireless and Cable Research group. She is working on measuring and modeling the propagation of electromagnetics fields around wireless systems. She is also doing measurements of electromagnetic fields around base stations in order to check their compliance with the exposure limits. Emmeric Tanghe was born in Tielt, Belgium on August 31, He received the M.Sc. degree in electrical engineering from Ghent University (Belgium) in July Since September 2005, he is a research assistant at IBBT-Ugent/INTEC (Interdisciplinary institute for BroadBand Technology-Department of Information Technology) of the same university. His scientific work is focused on the modeling of indoor and outdoor propagation through field measurements. Luc Martens (M 92) was born in Gent, Belgium on May 14, He received the M.Sc. degree in electrical engineering from Ghent University (Belgium) in July From September 1986 to December 1990 he was a research assistant at the Department of Information Technology (INTEC) of the same university. During this period, his scientific work was focused on the physical aspects of hyperthermic cancer therapy. His research work dealt with electromagnetic and thermal modeling and with the development of measurement systems for that application. This work led to a Ph.D. degree in December Since January 1991, he is a member of the permanent staff of the Interuniversity MicroElectronics Centre (IMEC), Ghent, and is responsible for the research on experimental characterization of the physical layer of telecommunication systems at INTEC. His group also studies topics related to the health effects of wireless communication devices. Since April 1993 he is Professor in electrical applications of electromagnetism at Ghent University.

9 JOSEPH et al.: OPTIMIZING COVERAGE AND THROUGHPUT FOR DVB-H SYSTEM BASED ON FIELD TRIALS 355 Etienne Deventer was born in Ronse, Belgium on February He graduated in the M. Sc. Degree in electrical engineering from the Universiteit Gent (Belgium) in Since 1974, he worked at the transmission department of VRT (public radio and tv company in Flanders). He was involved in frequency management, new wireless broadcasting technologies and project management. Until the end of 2007, he was a senior research expert at Medialab, the research department of VRT. Now he is acting manager of the transmission department of VRT. Hugo Gauderis was born in Vilvoorde, Belgium on February 24, He received the M.Sc. degree in electrical engineering from the Katholieke Universiteit Leuven (Belgium) in July He joined the transmission department of the Flemish public broadcaster in Belgium (VRT) in January 1993, where he worked on the new broadcast transmission technologies, frequency management and transmitter network planning for analogue and digital broadcasting. Since 2007, he works as senior expert in media distribution for VRT Medialab, which is the technological research department of the VRT. VRT Medialab carries out research into the creation, management and distribution of media content.