TERRESTRIAL television broadcasting has been widely

IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 245 A General SFN Structure With Transmit Diversity for TDS-OFDM System Jian-Tao Wang, Jian Song, Jun Wang, Chang-Yong Pan, Zhi-Xing Yang, Lin Yang Abstract In the digital television terrestrial broadcasting, the Single Frequency Network (SFN) has much advantage. SFN can serve an arbitrary large area with the same information broadcasted at the same frequency, resulting in the potential diversity gain. To improve the transmission performance of the Time Domain Synchronous-Orthogonal Frequency Division Multiplexing (TDS-OFDM) system in SFN, a general SFN structure with transmit diversity is introduced, theoretically analyzed computer simulated. The proposed SFN structure is flexible to set up. Simulations show that this method can greatly improve the reliability of the signal transmission over the frequency selective fading channels is suitable for the TDS-OFDM system in SFN. Index Terms Digital television terrestrial broadcasting (DTTB), single frequency network (SFN), space-time block code (STBC), terrestrial digital multimedia/television broadcasting (DMB-T), time domain synchronous-orthogonal frequency division multiplexing (TDS-OFDM), transmit diversity. I. INTRODUCTION TERRESTRIAL television broadcasting has been widely used, especially in those countries with very large area. The traditional analog television transmitters usually broadcast to the adjacent areas with different RF channels to avoid the so-called artificial multi-path propagation as the viewers are extremely sensitive to the in-b interference. This is called Multi-Frequency Network (MFN) which has inefficient radio spectrum utilization. With the rapid development of the digital television terrestrial broadcasting (DTTB), a special network structure called Single Frequency Network (SFN) is proposed. In SFN, all the transmitters broadcast the identical information simultaneously at the same frequency. High spectrum efficiency is therefore, the major advantage of the SFN [1], [23]. In SFN, the received signal can be seen as a mixture of multiple delays of the transmitted signal [2]. This observation leads to an equivalent convolutional channel between the transmitted received signal. With the SFN structure, as every transmitter emits the same signal at the same frequency, it is quite often that the received signal has a large delay spread, which is quite different from the scenario under MFN structure. In the traditional SFN scheme, the time dispersion, including the so-called artificial multi-path propagation caused by the SFN structure the natural propagation caused by the reflection in the vicinity of the receiver, are usually mitigated by equalization Manuscript received October 17, 2005; revised January 19, 2006. This work was supported in part by the China National Science Foundation under Grant 50177001 60372007 by the Ministry of Information Industry Foundation under Grant 2002291. J.-T. Wang, J. Wang, C.-Y. Pan, Z.-X. Yang, L. Yang are with the Department of Electronic Engineering, Tsinghua University, Beijing 100084, China (e-mail: wjt01@mails.tsinghua.edu.cn). J. Song is with the Research Institute of Information Technology, Tsinghua University, Beijing 100084, China (e-mail: jsong@tsinghua.edu.cn). Digital Object Identifier 10.1109/TBC.2006.872988 multi-carrier scheme [3], [4]. However, if some of the transmitters are located too far away from the receiver, the delays of the signals from those transmitters will be too large the traditional methods mitigating the interference will pay high cost. So it still faces the tough challenge of the strong echoes of very large delay spreads due to the signal transmission from the different transmitters within SFN. Considering the received signal is a superposition of signals coming from quite a few transmitters in SFN, if one or several transmitters are shadowed, others are still receivable. This merit results in the possible diversity gain. Over the fading channels, spatial diversity technique can effectively improve the receiving performance of the system [5]. It has been implemented at the receiver end with the request of the multiple RF front-end circuits high complexity [6]. This complexity issue is a major drawback for the portable or mobile receivers where physical size power consumption are the most important constrains. Comparing with the receiver diversity approach, transmit diversity scheme achieves diversity gain without greatly increasing the complexity of the receivers is more suitable cost effective for the broadcasting networks. So the transmit diversity technique has received strong interests in recent years [7] [12]. A simple yet efficient SFN structure with transmit diversity for the Time Domain Synchronous-Orthogonal Frequency Division Multiplexing (TDS-OFDM) system is proposed analyzed in this paper, showing the significant performance improvement. Based on the frame structure of the TDS-OFDM system, the proposed SFN structure is flexible to set up the computation complexity for the transmitter is very small, almost the same as that of the single antenna system. The rest of this paper is organized as follows. Section II describes the system model of TDS-OFDM the equivalent channel express for SFN. As an effective transmit diversity approach, space-time block code (STBC) is also briefly reviewed. In Section III, the general SFN structure for TDS-OFDM system is introduced. Section IV gives the analysis of the system performance computational complexity. Bit error rate (BER) simulation results demonstrating the performance are provided in Section V. Finally, some conclusions are drawn in Section VI. II. SYSTEM DESCRIPTION A. TDS-OFDM System TDS-OFDM is the modulation scheme for the Terrestrial Digital Multimedia/Television Broadcasting (DMB-T) system. DMB-T system is proposed by Tsinghua University as one of the major cidates for Chinese DTTB stard [13]. The transmitter modulates data with inverse discrete Fourier transform (IDFT), which is the same as in DVB-T stard [14]. However, pseudorom noise (PN) sequences are inserted into the header of the frame as the guard intervals as well as 0018-9316/$20.00 2006 IEEE 转载

246 IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 Fig. 1. The structure of signal frame in TDS-OFDM system. the training symbols for the channel estimation the timing recovery. At the receiver end, TDS-OFDM is equivalent to the zero-padded OFDM (ZP-OFDM) after removing the PN sequences. More research results on TDS-OFDM system can be referred to [13], [15] [20] the literatures therein. In the TDS-OFDM system, a signal frame consists of the data (frame body) the guard interval (frame header). There are 3780 symbols in the frame body which is always 500 long. So the frequency interval between the adjacent sub-carriers is 2 khz. The length of guard interval may be 1/9 or 1/4 of the length of the frame body. Fig. 1 shows the structure of the signal frame in TDS-OFDM system. Fig. 2. The simplified SFN structure. B. Equivalent Channel Express for SFN In SFN, the received signal is the mixture of multiple delays of the transmitted signals. A simplified SFN system is shown in Fig. 2, where the transmitted signals arrive at the receiver along four directions. It is noted that the natural propagation paths are omitted in the simplified figure. A field testing result of the impulse response in a SFN channel is shown in Fig. 3. From the figure, we can observe two path clusters, i.e. the receiver can receive the enough strong signals from two transmitters in the network these transmitters are called contributing transmitters in the following. So, each cluster in the figure corresponds to a contributing transmitter. There is always a main path in each cluster, representing the direct path of each transmitter/receiver pair. The other paths within the cluster st for the local scattering. It can be seen from Fig. 3 that large delay spread strong echoes exist in the equivalent channel, introducing the serious frequency selective fading. Later on in the next section, we will demonstrate that large delay spread strong echoes will not appear if the transmit diversity technique is applied. Because the channel state information (CSI) for each transmitter can be obtained through certain arrangement, that is, the channel condition of each transmitter/receiver pair is identified. The whole SFN channel can then be decomposed into several equivalent channels corresponding to each transmitter/receiver pair with much smaller delay spread. C. Review of STBCs Recently, transmitter diversity has been extensively studied as one of the promising approach to combat the detrimental effects in the wireless fading channels. With the very simple maximum-likelihood (ML) decoding algorithm implemented at the receiver, space-time block coding scheme has are still receiving much attention as an effective method to achieve the transmit diversity. STBC from the orthogonal designs is first proposed by Alamouti [7], then generalized by Tarokh, Fig. 3. An example of impulse response in SFN channel. Jafarkhani, Calderbank [8]. The main property of these codes is the orthogonality between each pair of the columns in the transmission matrices. Thus, the transmitted symbols can be decoded separately, not jointly. The orthogonal codes can achieve full diversity, but it is impossible to achieve the full transmission rate (i.e. rate 1) when there are more than two transmit antennas [8]. To get rate 1 codes, the STBC s from quasiorthogonal designs have been proposed [11], [12] with the columns of the transmission matrix divided into groups. The columns within the same group are not orthogonal to each other, while the columns from different groups are orthogonal to each other. The ML decoding can still be performed by searching for the symbol-pairs in each group with the complexity reasonably small. The OFDM systems can also employ STBC method to obtain the better signal quality from the diversity gain [9], [10]. The proposed scheme in the next section is based on the STBC method. III. THE GENERAL SFN STRUCTURE WITH TRANSMIT DIVERSITY FOR TDS-OFDM SYSTEM Generally, considering the SFN planning, we assume that there are no more than four contributing transmitters received anywhere in the network. The signals from other transmitters in the network have greatly faded for the long distance away from the receiver. In the following, we describe the SFN structure analyze the system performance in detail according to

IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 247 The transmitter of the second branch is denoted as Tx2. In the space-frequency encoder module, we perform the block coding across the two adjacent sub-carriers to get the following signals, (3) Then the time domain signal for Tx2 can be constructed as Fig. 4. The diagram of the proposed SFN structure with transmit diversity for TDS-OFDM system. According to the property of DFT, the results of DFT of may be written as (4) -point the number of the contributing transmitters in the network, respectively. The diagram of the introduced SFN structure with transmit diversity is shown in Fig. 4. Because multiple transmitters are used at the same frequency at the same time, each transmitter should use a unique PN sequence, orthogonal to each other. It is assumed that different transmit branches are statistically independent the perfect CSI is obtained at the receiver. In the following, a complex conjugation, transposition Hermitian operator are denoted as the superscript,,, respectively. denotes the congruent number of with modulus, that is. We consider the TDS-OFDM system with sub-carriers. Let represent the frequency domain input sequence after symbol mapping denote the symbol of the th sub-carrier in the th frame. are half length vectors denoting the even odd component of the vector, respectively. A. Two Contributing Transmitters If there are at most two contributing transmitters in SFN, only the modules outside the dashed polygon exist in Fig. 4. At first, the input bit stream is mapped into the symbol stream according to the modulation scheme. Then, the time domain signal after the inverse discrete Fourier transform (IDFT) is given by (in this subsection, is omitted for notational simplicity) So the equivalent signal configurations of the two transmit branches are given by At the receiver, the symbol-wise ML decoding can be applied to, similarly to [7]. So this method achieves the 2-order transmit diversity. The detailed description of the decoding procedure is given in the Appendix. B. Three Four Contributing Transmitters First, we discuss the case with four contributing transmitters in SFN. The modules inside the dashed polygon are included in Fig. 4. The time domain signals for Tx1 Tx2 are the same as that in Section III-A. However, the -point IDFT results of the consecutive two frames, i.e.,,, are all stored in the memory. The space-time encoder module performs the block coding across the two adjacent OFDM frames. For Tx3, we get (5) (6) (1) where. Let, which are stored in the memory, denote the results of -point IDFT of, respectively. Then from (1), the time domain signal for the first antenna Tx1 can be rewritten as (2) (7) Then it is similar to (2), the time domain signals for the third antenna Tx3, i.e., can be obtained. Again, according to the characteristic of DFT, the results of -point DFT of,, are,,, respectively. So the equivalent consecutive inputs of the OFDM frames for Tx3 are (8)

248 IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 As shown in Fig. 4, the signals for the fourth antenna Tx4 are based on,, through the space-time encoder module. (9) Also similar to (2), the time domain signals for Tx4, i.e., can be achieved. The results of -point DFT of,,, i.e.,,, are (10) From (6), (8) (10), we can observe that the equivalent transmitted signal matrix of the proposed scheme is shown in the equation at bottom of page. It is easy to prove that is a quasiorthogonal STBC matrix with four transmit antennas [11]. So the symbol pairs of can be decoded separately the diversity is 2. Again, the detailed description of the decoding procedure is also given in the Appendix. In order to achieve full diversity, i.e. the diversity order is 4, we introduce the phase-shift module before the space-frequency encoder in the proposed structure as shown in Fig. 4 with dotted line. Referring to [12], the optimal rotation angle is. For the case with three contributing transmitters in SFN, we can only simply take out the fourth transmit branch Tx4. The quasiorthogonal structure is still valid for the rest of three branches [11], i.e. Tx1, Tx2 Tx3. The diversity order is still 2 without the constellation rotation (CR) module. IV. SYSTEM ANALYSIS A. Performance Because each transmitter uses a unique PN sequence as the guard interval (GI) of TDS-OFDM system, orthogonal to each other, the CSI of each transmitter/receiver pair can be identified at the receiver end. Then, the whole SFN channel can be decomposed into several equivalent channels corresponding to each transmitter/receiver pair with much smaller delay spread. Meanwhile, the potential diversity gain is achieved. From the analysis in the Appendix, it is observed that the introduced transmit diversity method is similar to that of the maximal ratio combining (MRC) receiver diversity system [21]. In the following, we use the case with two contributing transmitters as an example to state how the diversity gain is obtained. In order to compare the performance of one-transmitter system with two-transmitter system fairly, the transmission power for each transmitter in the two-transmitter case is halved so that the average received signal powers are the same in both cases. The signal to noise ratio (SNR) on the th sub-carrier is given by (referring to () in the Appendix) (12) where represents the signal to noise ratio of every symbol, assuming the same modulation scheme is applied to all the sub-carriers. Suppose that the different echoes in the multi-path channel are wide sensed stationary uncorrelated complex Gaussian process the channel response are both zero-mean complex Gaussian rom variables with variance of 1. So the rom variable satisfies the central -distribution whose degree of freedom is 4. The variance of every degree is 1/2. Then the mean variance of can be expressed as While in the one-transmitter system, the SNR is given by (13) (14) (15) The rom variable is also the central -distribution with degree of freedom equal to 2. The variance of every degree is 1/2. Then the mean variance of will be (16) (17) Comparing with the equations of (13) to (14) with (16) to (17), the mean of SNR doesn t change with the increase of number of transmitters within the network. This is different (11)

IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 249 from the receiver diversity system. However, the variance of SNR in the two-transmitter system is half of that in the onetransmitter system. This is an excellent example of how the transmit diversity method can smooth the channel fading improve the system performance. Moreover, in all the cases mentioned in Section III, if one transmit branch doesn t work, i.e., the corresponding terms of is zero, the ML decision can still work. So the introduced SFN structure with the transmit diversity method can also support Soft Failure. TABLE I TYPICAL DTTB CHANNEL MODEL B. Complexity We consider the more complicated case, i.e. the case with four contributing transmitters in SFN. From the signal configurations given by (3), (7) (9), based on the symmetric property of DFT/IDFT, the time domain signals for Tx2, Tx3, Tx4 could be generated by the simple computations on,,, which have already been obtained through the IDFT operation for Tx1. Therefore, to obtain all the transmitted signals for the four transmit antennas, there will be one -point IDFT operation, additional complex multiplications complex additional additions required for each OFDM frame time slot. While, if the straightforward implementation of transmit diversity for OFDM scheme is adopted [9], four -point IDFT operations are needed within one OFDM frame time slot in four-antenna case. So comparing with the straightforward implementation, the computational complexity of the transmitter is greatly reduced. In the proposed scheme, each transmitter uses a unique PN sequence as the frame header. Then at the receiver end, we can use the multiple known PN sequences to perform the channel estimation time synchronization by means of the shift correlation [13], [15], [18]. Since the PN sequences for the different transmitters are orthogonal to each other, the channels corresponding to each transmitter/receiver pair can be easily separated, i.e. the whole SFN channel is decomposed into several equivalent channels. So the complexity at the receiver end is only linearly increased comparing with those scheme without the transmit diversity. Moreover, since the proposed SFN structure adopts a hierarchical diversity form as shown in Fig. 4, the network is flexible to set up. V. SIMULATIONS The BER performance of the introduced SFN scheme with the transmit diversity for TDS-OFDM system was further verified by simulations. This channel model is romly picked from the Brazil digital television test report [22] for the simulations with the parameters listed in Table I. It is a typical frequency selective fading channel model in DTTB. We only use this channel model as an example to demonstrate the effectiveness the proposed scheme can perform well under other channel conditions. The parameters of the TDS-OFDM system for simulations are listed in Table II. In our simulations, it was assumed that the transmit power of all the transmitters in SFN were equal, the fading paths from TABLE II THE PARAMETERS OF TDS-OFDM SYSTEM each transmitter to the receiver were independent, the CSI of each transmit branch was perfectly estimated at the receiver. Although in the analysis of Section III Appendix, some channel assumptions are made to meet the requirement of STBC structure, we still omit the assumptions in the simulations use Rayleigh model for each transmit branch to approximate to the practical channels. We now investigate the relationship between BER SNR for the portable reception [24], [25] in SFN environment. Simulation results for the system with QPSK, 16QAM 64QAM modulation on each sub-carrier are shown from Figs. 5 to 7, respectively. The maximum Doppler shift is set to 20 Hz, corresponding to the low receiver velocity of about 26 to 50 km/h in the TV UHF b (@470 862 MHz). The simulation results give the excellent performance to show how the proposed scheme achieves the potential diversity gain in SFN. We first consider the SFN structure with two contributing transmitters the BER of is used as the bench mark. It can be seen from the figures that the proposed scheme achieves no less than 9-dB diversity gain in all three modulation cases, comparing with the conventional SFN system without transmit diversity. Then, from the slopes of the BER curves in the figures, we can observe that due to the quasiorthogonal design structure, the three-antenna four-antenna cases have the same diversity order as that of the two-antenna case, only a little difference among them. But, if the constellation rotation method is adopted, the four-antenna case can obtain full diversity order [12], as shown by the BER curves with in the figures. Here in the simulations, the optimal rotation angle is set to. So the proposed SFN scheme can greatly improve the system performance over the frequency selective fading environment.

250 IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 Fig. 5. The BER performance comparison of the TDS-OFDM systems with without transmit diversity method with QPSK modulation on each sub-carrier. Fig. 7. The BER performance comparison of the TDS-OFDM systems with without transmit diversity method with 64QAM modulation on each sub-carrier. the hierarchical diversity structure shown in Fig. 4, the network is flexible to set up. APPENDIX In this appendix, we will provide the detailed descriptions on how the ML detection is performed at the receiver. A. Symbol Detection for Two-Antenna Case Let denote the complex channel gains from transmitter to the receiver denote complex noise. Assume that the channel responses between adjacent sub-carriers are approximately the same, i.e.,, 2. Then from (6), the received signal can be expressed as Fig. 6. The BER performance comparison of the TDS-OFDM systems with without transmit diversity method with 16QAM modulation on each sub-carrier. VI. CONCLUSIONS SFN is believed to be a most viable solution for the terrestrial digital TV broadcasting network with supreme performance of high spectrum efficiency. In SFN, the natural multi-path interferences are generally much smaller than the main signal its impact on the receiver can be easily eliminated. The strong artificial multi-path from SFN itself appears when the signals from several transmitters are received. To combat against this particular problem, an efficient SFN structure with the transmit diversity method is introduced. The performance improvement of TDS-OFDM system in SFN with this method is theoretically analyzed simulated in this paper, showing its superior performance over the frequency selective fading environment. The decomposition of the artificial multi-path from SFN is easily done with the proposed scheme. Other than mitigate this artificial multi-path issue, this method can also achieve the diversity gain support the soft failure. Moreover, due to Let denote the channel response matrix, i.e. Both sides of (7) are multiplied by given by where terms are, (18) (19) the result is (20) the complex noise Then from the decision (20), we can perform the ML detection on separately.

IEEE TRANSACTIONS ON BROADCASTING, VOL. 52, NO. 2, JUNE 2006 251 B. Symbol Detection for Four (or Three)-Antenna Case For the case with four contributing transmitters, it has been shown that the proposed method is a quasiorthogonal STBC scheme. Let denote the complex channel gain of the th sub-channel from the antenna to the receiver at the th time slot denote the complex Gaussian noise. Assume the channel responses between both two adjacent frames sub-carriers are approximately the same, i.e. (21) Then, the two consecutive OFDM frame signals at the receiver can be expressed as where Then left multiply denote the channel response matrix, i.e. (22) to the both sides of (15) we get (23) with complex Gaussian noise,. Then from (16), we can get the pair-wise ML detection, i.e. the symbol pairs of can be decoded separately. In the three-antenna case, the fourth transmit branch Tx4 in Fig. 4 is taken out. The decision (16) can still be used, only is zero. So in this case,. ACKNOWLEDGMENT The authors would like to thank the anonymous reviewers for their valuable comments. REFERENCES [1] G. Malmgren, On the performance of single frequency networks in correlated shadow fading, IEEE Trans. Broadcasting, vol. 43, no. 2, pp. 155 165, Jun. 1997. [2] P. Magniez A. Gorokhov, Space-time equalization for DVB-T in single frequency networks, in Proc. Thirty-Third Asilomar Conference on Signals, Systems Computers, vol. 1, Oct. 1999, pp. 622 626. [3] A. Ligeti J. Zer, Minimal cost coverage planning for single frequency networks, IEEE Trans. Broadcasting, vol. 45, no. 1, pp. 78 87, Mar. 1999. [4] P. Angueira, M. Velez, D. de la Vega, G. Prieto, D. Guerra et al., DTV reception quality field tests for portable outdoor reception in a single frequency network, IEEE Trans. Broadcasting, vol. 50, no. 1, pp. 42 48, Mar. 2004. [5] T. S. Rappaport, Wireless Communications. Englewood Cliffs, NJ: Prentice-Hall, 1996. [6] Mobile DVB-T using antenna diversity receivers, G. Faria. (2001). http://www.broadcastpapers.com [Online] [7] S. M. Alamouti, A simple transmit diversity technique for wireless communications, IEEE J. Select Areas in Communications, vol. 16, no. 8, pp. 1451 1458, Oct. 1998. [8] V. Tarokh, H. Jafarkhani, A. R. Calderbank, Space-time block codes from orthogonal designs, IEEE Trans. Inform. Theory, vol. 45, no. 5, pp. 1456 1467, Jul. 1999. [9] K. F. Lee D. B. Williams, A space-frequency transmitter diversity technique for OFDM systems, in Proc. IEEE GLOBECOM 00, San Francisco, 2000, pp. 1473 1477. [10] Y. Gong K. B. Letaief, An efficient space-frequency coded OFDM system for broadb wireless communications, IEEE Trans. Communications, vol. 51, no. 11, pp. 2019 2029, Nov. 2003. [11] H. Jafarkhani, A quasiorthogonal space-time block code, IEEE Trans. Commun., vol. 49, pp. 1 4, Jan. 2001. [12] W. Su X.-G. Xia, Signal constellations for quasiorthogonal spacetime block codes with full diversity, IEEE Trans. Inform. Theory, vol. 50, no. 10, pp. 2331 2347, Oct. 2004. [13] Terrestrial Digital Multimedia/Television Broadcasting System, Patent 00 123 597.4. [14] Digital Video Broadcasting (DVB); Framing Structure, Channel Coding Modulation for Digital Terrestrial Television,, Tech. Rep. EN 300 744 v1.1.2, ETSI, Aug. 1997. [15] J. Wang, Z.-X. Yang, C.-Y. Pan, M. Han, L. Yang, A combined code acquisition symbol timing recovery method for TDS-OFDM, IEEE Trans. Broadcasting, vol. 49, no. 3, pp. 304 308, Sep. 2003. [16] Z. Yang, L. Tong, L. Yang, Outage probability comparison of CP-OFDM TDS-OFDM for broadcast channels, in Proc. IEEE GLOBECOM 02, Nov. 2002, pp. 594 598. [17] Z.-W. Zheng, Z.-X. Yang, C.-Y. Pan, Y.-S. Zhu, Novel synchronization for TDS-OFDM-Based digital television terrestrial broadcast systems, IEEE Trans. Broadcasting, vol. 50, no. 2, pp. 148 153, Jun. 2004. [18] Z. W. Zheng, Z. X. Yang, C. Y. Pan, Y. S. Zhu, Synchronization channel estimation for TDS-OFDM systems, in Proc. IEEE VTC 2003-Fall, Oct. 2003, pp. 1229 1233. [19] Z.-X. Yang, M. Han, C.-Y. Pan et al., A coding modulation scheme for HDTV services in DMB-T, IEEE Trans. Broadcasting, vol. 50, no. 1, pp. 26 31, Mar. 2004. [20] Z.-X. Yang, Y.-P. Hu, C.-Y. Pan, L. Yang, Design of a 3780-Point IFFT processor for TDS-OFDM, IEEE Trans. Broadcasting, vol. 48, no. 1, pp. 57 61, Mar. 2002. [21] T. Eng, N. Kong, L. B. Milstein, Comparison of diversity combining techniques for Rayleigh-fading channels, IEEE Trans. Communications, vol. 44, no. 9, pp. 1117 1129, Sep. 1996. [22] Digital Television Systems Brazilian Tests Final Report, F. Pollara.http://www.set.com.br/testing.pdf [Online] [23] A. Mattsson, Single frequency networks in DTV, IEEE Trans. Broadcasting, vol. 51, no. 4, pp. 413 422, Dec. 2005. [24] Implementation Guidelines for DVB Terrestrial Services: Transmission Aspects, ETSI, Tech. Rep. TR 101 190 v1.1.1, Dec. 1997. [25] P. Angueira, M. Velez, D. de la Vega, A. Arrinda et al., DTV (COFDM) SFN signal variation field tests in urban environments for portable outdoor reception, IEEE Trans. Broadcasting, vol. 49, no. 1, pp. 81 86, Mar. 2003.