Analysis and Performance Comparison of DVB-T and DTMB Systems for Terrestrial Digital TV

Analysis and Performance Co mparison of DVB-T and DTMB Systems for Terrestrial Digital TV Ming Liu, Matthieu Crussière member IEEE , Jean-François Hélard senior IEEE , Oudomsack Pierre Pasquero Institute of Electronics and Telecomm unications of Rennes (IETR) Rennes, France {first name. last nam e}@insa-rennes.fr Abstract — Orthogonal f requency-division multiplexing (OF DM) is the most popular transmission technolog y in digital terrestrial broadcasting (DTTB), adopted by many DTTB standards. In this paper, the b it error rate (B ER) performance of t wo DTTB systems, namely cyclic prefix OFDM (CP-OFDM) based DVB-T and time domain synchronous OFDM (TDS-OFDM) based DTMB, is evaluated in dif ferent channel conditions. Spectrum utilization and po wer efficiency ar e also discussed to demonstrate the transmission overhead of b oth systems. Simulation results show that the performances of the two systems are much close. Given the same ratio of guard interval (GI), the DVB-T outperforms DTMB in terms of s i gnal to noise ratio (SNR) in Gaussian and Ricean channels, while DTMB behaves better performance in Ra yleigh channel in hi gher code rates and higher orders of constellation thanks to its efficient channel cod ing and interleaving scheme. Keywords-Digital TV; DVB-T; CP-OFDM; DTMB; TDS- OFDM I. I NTRODUCTI ON Transition from analog to digital television (DTV) is a trend worldwide. DTV services can be delivered via satellite, cable and terrestrial broadcasting. Due to its flexibility to both stationary and mobile ap plications, digital ter restrial television broadcasting (DTTB) has attracted more and more interest in recent years. Nowadays, there are three main DTTB standards around the world: Digital Video Broadcasting-Terr estrial (DVB-T) [1] in Europe, the tr ellis-coded 8-level vestigial side band ( 8-VS B) modula tion s yst em de velo ped by Adva nced Television System Committee (ATSC) [2] in North America, the Integrated Services Digital Broadcasting-Terr estrial (ISDB- T) [3] in Japan. Among the m, DVB-T plays the most importan t role. Since first approved in 1997, DVB-T has become the dominant terrestrial broadcast ing standard in Europe and is al so popular in other continents. By J une 2008, DVB-T services have been launched by 33 countries and territories. After 12 years of developing, the Chinese Digital Terrestrial/Television Multimedia Broad casting (DTMB) standard [4] was finally ratified in August 2006, and began t o be a mandatory national s tandard in August 2007. D TMB consists of single carrier modulation (C = 1) and multicarrier modulation (C = 3780) which are originate d from two former proposals: the single-carr ier ADBT-T (Advanced Di gital Television Broadcasting-Ter restrial) and the multi-c arrier DMB-T (Digital Multimedia/TV Broadc asting-Terrestrial) respectively, providing flexible combinations of working modes for different application scenar ios. Because of the enormous TV market in China and the nove l signal processing techniques integrated in it, the Ch inese DTMB draws great interests from both industries and researchers . The orthogonal frequency-div ision multiplexing (OFDM) is definitely the most popular technique adopted by majority of DTTB standards (DVB-T, ISDB-T and DTMB). This is due to its robustness to frequency select ive fading. By implementing inverse fast Fourier transform (IFFT) and FFT at transmitter and receiver sides respectively, OFDM transforms a high speed serial data flow to a set of low speed parallel ones at ort hogonal flat fading sub-channels. Traditionally, a cyclic pre fix (CP) is inserted between two consecutive OFDM sy mbols as guard interval (GI). This solution has been chosen for many standards, namely for DVB-T and ISDB-T. The length of the GI is designed to be longer than that of channel memory. By discarding the CP at the receiver, the inter sy mbol interference (ISI) is then re moved from the received signal. With the as sistance of CP, the linear convolution between trans mitted signal and channel impulse response (CIR) converts int o a circular one i.e. the channel convolution effect is t urned to be a set o f parallel attenuations in the discrete frequency domain. Hence, the equalizatio n in OFDM can be perfor med by simply multiply ing a coefficient on each subcarrier at the receiver. T hus, the equalization complexity of OFDM is significantly low compared with a time domain equalizer. As the samples for the CP do not convey useful data, several researchers prop osed to replace the CP by known pseudo noise (PN) sequences . This becomes the time domain synchronous OFDM (TDS- OFDM [5], also known as pseudo random postfix OFDM, PRP -OFDM [6] and known symbol padding OFDM, KSP-OFDM [7]). Besides serving as GI, the PN sequence can also be exploited to make channel estimation and synchronization in the time domain. Hence, it is not necessary to insert scattered and continual pilots to the OFDM symbols, which increases the spectrum efficiency. Moreover, since the channel estimation can be performed for each OFDM sym bol in th e tim e domain , TDS- OFDM can a chieve a f ast channel acquisition. However, in contrast with the CP-OFDM, the circularity property mentioned above is no longer obtained and specific algorithms have to be processed at the receiver to This work is supported by the French national project “Mobile TV World ”. restore the cyclicity of the signal. In particular, the PN sequence has to be perfectly removed before demodulation, thus leading to a signal referred to as zero padding OFDM (ZP- OFDM) which has been adopted, for instance, in the Wi Media solution for ult ra wide band (UWB) c ontext [9]. T he resulting signal can then be demodulated using the esti mation methods developed for ZP-OFDM, as proposed in [8] for exa mple. Chinese DTMB standard ( multicarrier mode) is so mewhat similar to DVB-T in terms of OFDM trans mission scheme, signal constellation, and 8 MHz analog b andwidth. Therefore DVB-T is an ideal counterpart of Chinese DTMB standar d to analyze the performance and make comparisons. Although there are some measured results presented in [10], [ 11] in manners of sensitivity and carrier -to-noise ratio (C/N), there is still not so much reference for DTMB as it fo r DVB-T. So, it is necessary to carry out a survey on DTMB and make a precise comparison between multicarrier mode of DTMB and DVB-T in order to provide a reference for researchers who may be interested in DTMB. Rest parts of this paper are organized as follows. Section II describes the main features of DVB-T and DTMB sy stems. A brief discussion on power factor of both systems is also made in this section. Simulation results for both Gauss ian channel and multipath channels are presented in Sect ion III. Conclusions are drawn in Section IV. II. P RESENTATI ON AND A NALYSIS OF DVB-T AND DTMB S YSTEMS A. DVB-T System DVB-T adopts standard CP-OFDM as trans mission scheme. The modulated data symbols are transmitted block-by- block. The i th data block x M ( i ) is an M × 1 com plex vector in the frequency domain whose elements are complex symbols coming from quadrature amplitude (QAM) modulation. After performing M -points IFFT, it yields the ti me domain block: () () H MM M ii = F % xx ( 1 ) wher e F M is an M × M FFT mat rix with ( m , k )th entry M − (1/2) exp{ − j2 π nk/M }, (.) H denotes Hermitian transposit ion, the subscript M indicates its size of either an M × 1 vector or an M × M m atrix and the tilde ( ˜ ) denotes time domain variables. Then a CP of length D is inserted between two consecutive blocks. Let P = M + D be the total length of transmitted symbols per block, and let I cp = [ I c , I M ] T be the P × M matrix representing the CP appending, where I M stands for the siz e M identity matrix and I c is the M × D m atrix corresponding to the D last columns of I M . The i th block of symbols to be transmitted can be expressed as () () () H cp cp M cp M M ii i == II F %% xx x . (2) The multipath channel can be m odeled as an L th order FIR Figure 1. Block diagram of DVB-T receiver filter with impulse re sponse [ h 0 , …, h L -1 ]. Let H ISI and H IBI be the intra and inter block interference. H ISI and H IBI are P × P Toeplitz lower and upper triangular matrices with the first column [ h 0 , …, h L -1 , 0, …, 0 ] T and first row [0, …, 0, h L-1 ,…, h 1 ] respectively. The received i th block is: ISI IBI () () ( 1 ) () pc p c p p ii i i =+ − + HH %% % % rx x n . (3) Because OFDM system satisfies D ≥ L , the H IBI can be eliminated by removing the CP. And d ue to the circular structure of the CP-OFDM, H ISI turns to be an M × M circulant matrix H circ with the first row [ h 0 , 0, …, h L , …, h 1 ]. By the property that circulant matrix is diagonal in Fourier basis, after FFT, the received block becomes: () () () () () () ( ) ( ) MM c i r c c p M M H Mc i r c M M M M MM M M ii i ii M diag h i i = =+ =+ + FH F FH F F F %% % rx n xn xn ( 4 ) where diag ( . ) denotes a diagonal matrix with ele ments given by the vector argument, F M h M is the frequency response of the multipath channel. Thus, the transmitted signal x M can be easily recovered from r M by dividing a corresponding fading factor in the frequency domain. Fig. 1 presents the block diagram of DVB-T receiver. The GI is first removed from the received symbols. After FFT, pilots are extracted and channel es timation can be made based on them. Then, signals are equalized using the estimated channel frequency response in the frequency do main. The equalized data sy mbols are then converted to binary bits by demapper. Finally, the erroneous bits ar e corrected by channel coding combined w ith interle a ving . Table I gives the key parameters of DVB-T. There are four choices for the GI, providing a guard duration rang ing from 7 μ s to 56 μ s in 2K mode and 28 μ s to 224 μ s in 8K mode. Th ree constellations can be used with op tions of hierarchical modes. Both bitwise and sy mbol interleaving are perfor med to avoid long sequences of severely corrupted bits feeding to t he inner decoder of receiver, which can effectively improve the error correction ability of channel codi ng in presence of the frequency selective fading. The channe l coding consists o f Reed-Solomon RS (204, 118, t=8) and punctured convolutional code with code rate 1/2, 2/3, 3/4, 5/6 and 7/8. Between outer TABLE I. P ARAMETERS OF DVB -T Nb. of active subcarriers 1705 (2K mode), 6817 (8K mode) Length of GI (Fraction of useful data length) 1/4, 1/8, 1/16, 1/32 Mapping QPSK, 16QAM, 64QAM (optionally hierarchical) Coding Outer Reed-Solomon RS(204, 188, t=8) Inner Convolutional code with code rate 1/2, 2/3, 3/4, 5/6, 7/8 Interleaver Outer Convolutional interleaving Inner Bitwise + symbol interleaving and inner coding, a convolutional interleaver with a maximum delay of 2244 bytes is adopted. That means the data is convolutionally interleaved to spread burst errors at the output of the inner decoder over several OFDM blocks, while the bitwise and symbol interleaving are made within one OFDM block . B. DTMB System The TDS-OFDM wavefor m is selected as the basic transmission scheme for the multicarr ier mode of DTMB. The transmitted TDS-OFDM signal can be expressed as: () () H TDS z p M M P ii =+ IF xx c % ( 5 ) where I zp = [ I M , 0 M × D ] T which pads D rows o f zeros to the tai l of modulated signals, with 0 the null m atrix of the dimension given by the subscript. c P = [ 0 M × 1 , c D ] T fills th e padded zeros with a preselected PN sequence c D . Other variables are s imilar to the ones used for the description of DV B-T in the previous paragraph. From (5), after completely r emoving the PN sequence and its effects due to the channel memory, the received signal can be written in the form of a ZP-OFDM signal. Thus, a straight forward way to demodulate TDS- OFDM signal is to remove the PN sequence and use the demodulation algorithms developed for ZP -OFDM. Several methods are proposed with a variati on of performance and complexity in [8]. Within these methods, the overlap-add (OLA) algorithm is the least complex means at the expense of losin g channel-irre spec tive in vertibility. T he ZP -OFDM -OLA is derived from the fact that by splitting ZP- OFDM signal into upper M × 1 and l ower D × 1 part, then padding M – D rows of zeros to the latter part, the same M × M circulant CIR matrix as in (4) can be manually construc ted. This means that ZP- OFDM -OLA has iden tical eq ualization method a s CP-O FDM. Fig. 2 shows the block diagram of DTMB system with a PN-subtraction-OLA algorith m. At the receiver side, PN sequence convolved by the channel is first extracted with knowledge of CIR. Then, OLA operation is performed by copying the following GI and adding to the beginning par t of an OFDM symbol in order to compensate for the effect of GI on the OFDM sy mbol due to the channel memory, and to restore the orthogonality between subcarriers. A fter the OLA Figure 2. Block diagram o f DTMB receiver processing, ZP-OFDM has r oughly the same performance as CP-OFDM [7], [8]. The PN sequences are also used to make channel estimation and syn chronization, even i f the signal equalization is carried out in the freq uency domain like in DVB-T, as evident from Fig. 2. Table II presents the main parameters of DTMB by separately listing them in single and multi-carrier mode. It should be mentioned that it is a common “co mbination” of working modes as shown in [10] and [11], and does not mean that some parameters can only be used in specific mode. Actually, there exists other combinations such as PN595 + C=3780 as shown in [11]. By standard definition, three different PN sequences, PN4 20, PN945 and PN595, can be used as GI, while the PN 595 has different properties fr om others in terms o f average power, fixed pha se and generation method. DTMB supports five types of constellations: 4 QAM, 4QAM-NR, 16QAM, 32QAM and 64QAM. The time do main interleaver is the same as the outer interleaver in DVB-T, but with much longer interleaving dept h. The interleaving in DTMB is performed over a large number of OFDM blocks to obtain a high diversity gain. The frequency domain interleaving is only used in the multicarrier mode. The following frequency interleaving maps time domain inter leaved symbols to 3780 subcarriers in a scra mbling order. Concatena ted BCH and LDPC are selected as channel coding wit h three options of code rate. For the sake of comparison fairness, this paper only focus es on the mult icarri er mode of D TMB syst em. C. System Comparison In DVB-T, only the central 83% subcarrier s are actually available for data transmissi on. The remaining FFT points at side parts are deliberately shut do wn to limit the signal spectrum within 8 MHz analog bandwidth. In order to aid channel estimation and synchroniza tion, continual and scatt ered pilots are inse rted in OFDM sym bols, occupyin g more than 10% subcarriers. There are about 1% subcarriers allocated to the transmission parameter signali ng (TPS) which relates to the transmission para meters, e.g. channel coding and modulation. All these factors degrade the spectrum utilization in DVB-T and further introduce a useful data rate loss. On the other hand, in DTMB, the sy nchronization and channel estimation are perfor med by using PN sequences. So there is much less spectru m efficiency loss due to pilots. The only spectrum efficiency d egradation comes fro m the 36 symbols of system infor mation in each 3780-long OFDM block. These symbols take about 1% subcarriers, which is equivalent to the cost of TPS in DVB-T. Eventually, the spectrum utilization of DTMB is about 10% higher than that of DVB-T . TABLE II. P ARAMETERS OF DTMB Single carrier mo de Mult icarrier mode Origin Former ADTB-T Former DMB-T Number of subcarriers C = 1 C = 3780 PN sequence Frame Header Length 595 (1/6) 420 (1/4), 945 (1/9) Power Non boos t Boosted by 2 Phase Same in a superframe Different or same Mapping 4QAM-NR, 4QAM, 16 Q AM, 32 Q AM 4QAM, 16QAM, 64 Q AM Interleaver Time domain Time & Frequency domain Coding Outer BCH(762, 752) Inner LDPC(7493, 3048), (7493, 4572), (7493, 6096) Code rate 0.4(7488, 3008), 0.6(7488, 4512), 0.8 ( 7488, 6016 ) GI is also an expense of transmission power and useful data rate. In DVB-T, CP is a duplicate of data part with the sa me power. However, in DTMB, two types of PN sequence frame header are boosted to obtain better channel es timation and synchronization performance. The boosted PN sequence spend s more power than the non-boosted CP given the sa me GI length. In order to evaluate th e transmission cos ts in the two systems, all factors mentioned above should be take n into account. An evaluation of power efficiency can be obtained by calculating the ratio of the power allo cated to the data subcarriers over all power spent in the transmission. Specifically, in DVB-T, this power efficiency factor can be computed by 1 1 data DVB T data TPS pilot N N N N boost GI γ − × ++ × + = (6) where N data , N TPS , N pilot represent the number of data, TPS and pilot subcarrier, respectively. The boost is the boost factor for pilot subcarriers. The GI stands for the fraction of GI over dat a part. In DTMB, a si milar expression can be written as: 1 1 data DTMB data info N N N GI boost γ × ++ × = . (7) (7) uses the same notation as (6) except that N info represents the number of system information sy mbols. It should be noticed that this pow er efficienc y facto r prese nts th e power allocat ion in data subcarriers independent of mapping and coding scheme. Or, in other words, it is a measurement of the transmission overhead in terms of power, showing the efficiency of the system data structure. Parameter s and the resulting power efficiency factors are presented in Tab le III. In GI = 1/4 case, although DTMB has higher spectru m utilization ratio of 10%, DTMB and DVB-T have the same power efficiency factor . This is mainly due to the fact that the PN sequence which takes a large portion of transmitted signal, are booste d, decreasing the overall power efficiency. However, this problem is less TABLE III. P ARAMETER C OMPARISON B ETWEEN DVB-T AND DTMB DVB-T DTMB 2K mode 8 K mode C = 3780 FFT size 2 048 8 192 3 780 Nb. of subcarriers 1 705 6 817 3 780 Nb. of data subcarrie rs 1 512 6 0 48 3 744 Subca rrier s pacing 4 464 Hz 1 116 H z 2 000 Hz Signal bandwith 7.61 MHz 7.61 MHz 7.56 MHz OFDM symbol duration 224 µs 896 µs 500 µs Power effici ency factor 0.66 (GI=1/4), 0.73 (GI=1/8), 0.77 (GI=1/16), 0.79 (GI=1/32) 0.66 (GI=1/4), 0.81 (GI=1/9) significant in the GI = 1/9 mode of D TMB. The power efficiency factor in this cas e is 0.81, which is not only significantly higher than the equivalen t GI = 1/8 mode in DVB- T, but is even slightly higher than it in the case of the minimum GI of 1/32. Besides increasing the spectru m utilization, PN sequence makes it possible to achieve a faster ch annel acquisition in DTMB. In the DVB-T system, complete channel esti mation can be performed by using pilots from four consecutive OFDM blocks, while, in DTMB, it can be made for every block relying on its own PN sequence. T his feature is expected to make DTMB more robust in high mobility scenar io. III. S IMULATION R ESULTS In this section, we analyze the bit err or rate (BER) performance of the two systems in additive white Gaussian noise (AWGN), Ricean (F 1 ) and Rayleigh (P 1 ) channels. The latter two channels are specified in [1]. The length of GI is set to 1/4, because it is the only common op tion in both sys tems. Using the same ratio of GI will introduce identical performance loss due to the time domain redundancy which plays an important role in combating multipath effec ts. Another way to guarantee the fairness of co mparison is to find a pair o f working modes with a pproximately the sa me useful bitrate. In Table IV, three working modes are picked from each system, representing low, medium and high throughput applications, TABLE IV. S IMULATION P ARAMETERS AND U SEFUL B IT R ATES AT GI =1/4 Mode System Mapping Code Rate Bitrate (Mbps) No outer code With outer code No outer code With outer code 1 DVB-T QPSK 1/2 0.46 5.4 4.98 2 16QAM 3/4 0.69 16.2 14.93 3 64QAM 3/4 0.69 24.3 22.39 4 DTMB QPSK ≈ 0.4 ≈ 0.4 4.88 4.81 5 16QA M ≈ 0.6 ≈ 0.6 14.63 14.44 6 64QA M ≈ 0.6 ≈ 0.6 21.96 21.66 Figure 3. BER compar ison of DVB-T and DTMB in AWGN channel without outer code, GI = 1/4 respectively. Corresponding use ful bitrates are also given. The 2K mode is selected as the representa tive of DVB-T. In DTMB, the interleaving depth is c hosen as 240. The LDPC decoder adopts the message-passing algorith m with a maximum iteration times of 50 which is a good trade-o ff between error correcting perfor mance and time consuming. All the simulation results are presented in terms of BER versus signal to noise ratio (SNR) which is defined by the average signal power over noise power. A. In AWGN Channel Fig. 3 gives the comparison results witho ut taking into account the outer code in AWGN channel. In DVB-T, for quasi-error-free (QEF) reception, the projected post-RS BE R is less t han 10 − 11 , requiring a post-Viterbi BER of less than 2 × 10 − 4 which is taken for evalu ation here. C/N references g iven by [1] are also printed as cross in the figure. One has to be careful about the fact that these C/N re ferences actually correspond to the ratio of the power on data subcarriers over the power of noise, without considering the power o f pilots and GI as well as the inactive s ubcarriers. However, the SNR u sed here is the exact ratio of the received po wer inclu ding the power spent for pilot subcarriers over the power of noise. For this reason, the averaged SNR is sligh tly smaller than C/N. Specifically in DVB-T, the re ferences given in C/N should be shifted 0.46 dB to the left to get the corresponding re ferences in SNR. After this shift, it can be observed that si mulated curves of mode 1 to 3 exactly pass over t he references, proving the correctness of our simulations. The BE R performance of DTMB without outer code is also shown in the same figure. It can be seen that in ter ms of SNR, the two system s have alm ost the same performance in low and medium data rate cases, while in the high data rate situat ion, DTMB is 0.4 dB better when BER equals to 2 × 10 − 4 . Fig. 4 shows the simulation results with out er code in both systems. Profiting from the RS (204, 118) with 8 byte error correction capability and the interleaving between inner and Figure 4. BER comparis on of DVB-T and DTMB in AWGN channel with outer code, GI = 1/4 outer code, the performance of DVB-T is significantly improved, exhibiting a sharp flop. However , the BCH (762, 752) code of the DTMB syste m can only correct one bit error and does not exhibit any effect at BER level of 10 − 4 . No improvement can be observed when the outer code is added to the simulation in DTMB. From Fig. 4, DVB-T is 1.2 dB, 1.1 dB an d 1.0 dB better tha n DTMB at BER=5 × 10 − 5 in thre e mode s. Comparing Fig. 3 and Fig. 4, we can see different philosophies for the two systems. In DVB-T, the task of error correction is shared by inner and outer coding. So each o f them should be sufficiently e ffective and the interleaving be tween them is necessary. On the other hand, in DTMB, the duty o f forward error correcting is mainly fu lfilled by the LDPC code. The LDP C co de h as suc h su perior per for mance t hat th e ma jor role of BCH is actually to adapt the data frame lengths [11]. So we can understand why the interleaving proc ess between inner and outer code are omitted in DTMB. It is more reas onable to compare the two systems with full error cor recting ability. By under standing thi s, reset compa rison s are car ried out only in the “with outer code” case. B. In Multipath Channels Fig. 5 and Fig. 6 give the si mulation results in Rayleigh channel and Ricean channel, respectively. The BER is measured at the output of outer decoder for each system. All simulations are carried out under the assumption of perfect channel estimation and synchronization. In the P 1 channel, DVB-T is 0.7 dB better than DTMB at BER = 5 ×10 − 5 in the low throughput case, and this difference decreases at lower BER level. Moreover, DTMB outperfor ms DVB-T in the other two cases at the same BER and a greater difference ca n be foreseen at even lower BER level. In the F 1 Ricean channel, the presence of a line of s ight path makes the performance close to that obtained in AW GN. Hence, a similar conclusion can be made about the results in F 1 as in AWGN, though some degradations of DVB-T can be noticed at higher code rates and constellations. 0 2 4 6 8 10 12 14 16 18 20 22 24 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 S NR ( dB ) BER D V B -T, 4.98M bps D V B-T, 14.93M bps D V B-T, 22.39M bps D TM B, 4.81Mbps D TM B, 14.44Mbps D TM B, 21.66Mbps 0 2 4 6 8 10 12 14 16 18 20 22 24 10 -4 10 -3 10 -2 10 -1 10 0 S NR ( d B ) BER D V B- T, 5.4M bps D V B- T, 16.2M bps D V B- T, 24.3M bps DV B- T Re f e r en c e D TM B, 4.88M bps D TM B, 14.63M bps D TM B, 21.96M bps Figur e 5. BER co mparison of DV B-T and DT MB in P 1 channel with outer code, GI = 1/4 From Fig. 5, Fig. 6 and compar ing with the results in AWGN, the performance degradation in P 1 and F 1 channels is stronger in DVB-T. This can be understood by the fact that DTMB exhibits a better adaptability to the multipath channel than DVB-T, thanks to the LDPC code co mbined with the extremely deep interleaving. IV. C ONCLUSION In this paper, two DTTB standards － DVB-T and DTMB, based on different GI approaches, are presented, compared and analyzed. Discussions on power utilizatio n are taken to analyze the transmission overhead cost in power perspective. Power efficiency factors de monstrate that DTMB has equivalent power efficiency as DVB-T in long GI (1/ 4) case, while the short GI (1/9) case of DTMB has a better power efficiency compared with all cases in DVB-T. Simulati on results show that the performances o f two systems are really c lose. When the GI equals to 1/4, DVB-T seems to outperform DTMB in terms of BER versus SNR in AWGN and F 1 channels, while DTMB enjoys better performance in P 1 channel because its channel coding and interleaving sche me is more effective in strong fading environ ments. In further studies, the comparison of the systems will be carried out in time selective fadin g channels implementing real chan nel estimation algorithms. Figure 6. BER co mparison of DV B-T and DTMB in F 1 channel with outer code, GI = 1/4 R EFER ENCES [1] Digital video broadcasting (DVB); Framing structure, channel coding and modulation for terrestrial television, European Standard (EN) 300 744 V1.5.1, European Telecommunications Standards Institute (ETSI), Nov. 2004. [2] Advanced Television Systems Committee, “ATSC Digital Television Standard,” Doc ument A/53 , Sept. 19 95. [3] Association of Radio Industries and Businesses , “Terrestria l Integrated Services Digital Broadcasting (ISDB- T) — Specifications of Chann el Coding, Framing Structure, and Modulation,” Sept. 1998. [4] Framing Structure, Channel Coding and Modulation for Digital Television Terrestrial Broadcastin g System, Chinese National Standard GB 20600-2006. [5] J. Wang, Z. Ya ng, C. Pan, M. Han, L. Yang, “ A com bined code acquisition and symbol timing recovery method for TDS-OFDM,” Broadcasting , IEEE Transactions on, vol. 49, pp. 304-308, Sept. 2003. [6] M. Muck, M. de Courville, P. D uhamel, “A pseudorandom postfix OFDM modulator — semi-blind channel estimation a nd equalization,” Signal Processing, IEEE Transactions on, vol. 54, no.3, pp. 1005-1017, March 2006. [7] H. Steendam, M. Moenec laey, “Different Guard Interval Techniques for OFDM: Per formance Co mparison,” in Proc. Of MC-SS’07, Herrsching, Germany, May 7-9, 2007. [8] B. Muquet, Z. Wang, G.B. Giannakis, M. de Courville, P. Duhamel, “Cyclic prefixing or zero paddi ng for wireless multicarrier trans miss ions?,” Comm unicatio ns, IEE E Transa cti ons on, vol. 50, n o.12, pp. 2136-2148, Dec 2002. [9] A. Stephan, E. Guéguen, M. Cr ussière, J.-Y. Baudais , and J.-F. Hélard, “Optimization o f Linear Precoded OFDM for High-Data-Rate UWB Systems,” Wireless Communications and Networking, EURASIP Journal on, vol. 2008, Article ID 317257, 2008. [10] J. So ng, Z. Yang , L. Yang, K. Gong, C. Pan, J . Wang, Y. Wu, “Technical rev iew on Chinese Di gital Terrestrial Television Broadcasting Standard a nd Measurements on Some Working Modes,” Broadcasting, IEEE Transaction s on , vol.53 , no.1, pp.1 -7, March 20 07. [11] W. Zhang, Y. Guan, W. Liang, D. He, F. Ju, J. Sun, “An Introduction of the Chinese DTTB Standard and Analysis of the PN595 Working Modes, ” Broadcasting, IEEE Transactions on , vol.53, no.1, pp.8-13, March 2007. . 0 2 4 6 8 10 12 14 16 18 20 22 24 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 SNR ( dB) BER D V B- T,4.98M bps D V B- T,14.93M bps D V B- T,22.39M bps D TM B,4.81M bps D TM B,14.44M bps D TM B,21.66M bps 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 S NR ( d B ) BER D V B-T,4.98M bps D V B-T,14.93M bps D V B-T,22.39M bps D TM B,4.81M bps D TM B,14.44M bps