OFDM系統(tǒng)中一種有效的符號(hào)定時(shí)同步方案

1、2007年9 月 西安電子科技大學(xué)學(xué)報(bào)（自然科學(xué)版） Sep.2007 第34卷 增 刊 JOURNAL OF XIDIAN UNIVERSITY V ol.34 Sup.An eficient symbol timing synchronizationfor OFDM systemsGUO Yi, LIU Gang, GE Jian-hua(The National Key Lab. of Integrated Service Networks,Xidian Univ., Xian 710071, ChinaAbstract : In OFDM systems, the symbol tim

2、img error over multipath fading channels should stay inside theISI free range of the guard interval. For DAB systems, the correct symbol timing can be obtained using a2048-point IFFT of the demodulated phase reference symbol, but it demands heavy computational work andcomplex control circuits. In th

3、is paper, a new efficient symbol timing synchronization algorithm is proposed forDAB systems, which utilizes the good autocorrelation of the phase reference symbol in the time domain. Comparedwith the method associated with IFFT, the method proposed has a lower complexity and a nicer flexibility wit

4、h thesystem performance unchanged.Keywords : OFDM;Symbol Timing Synchronization;AutocorrelationOFDM 系統(tǒng)中一種有效的符號(hào)定時(shí)同步方案郭 漪, 劉 剛, 葛建華(西安電子科技大學(xué) 綜合業(yè)務(wù)網(wǎng)理論及關(guān)鍵技術(shù)國(guó)家重點(diǎn)實(shí)驗(yàn)室,陜西 西安 710071摘要:針對(duì)OFDM 系統(tǒng)中的定時(shí)同步問題,提出了一種適用于數(shù)字音頻廣播系統(tǒng)的符號(hào)定時(shí)同步方案,首先通過(guò)信號(hào)能量比值檢測(cè)實(shí)現(xiàn)粗定時(shí),然后利用時(shí)域相位參考信號(hào)良好的自相關(guān)性完成細(xì)定時(shí).與傳統(tǒng)方案相比,它能在保證系統(tǒng)詼諧性能的同時(shí),大大降低系統(tǒng)對(duì)緩存及控制部分的要

5、求,并且實(shí)現(xiàn)簡(jiǎn)單、靈活性高.關(guān)鍵詞：正交頻分復(fù)用；符號(hào)定時(shí)同步；自相關(guān)性中圖分類號(hào)：TN914.51 文獻(xiàn)標(biāo)識(shí)碼：A 文章編號(hào)： 1001-2400(2007S1-0009-05The research and development work on the orthogonal frequency division multiplexing (OFDM technique for high speed digital data transmission has received considerable attention and has made a great deal of progr

6、ess. As is well known, OFDM systems are sensitive to symbol timing offset. To obtain responsible demodulated signals, the symbol timimg error should stay inside the ISI free range of the guard interval1,2.The CIR(channel impulse response-based symbol synchronization for the Eureka-147 DAB system est

7、imates the CIR, or the exact multipath delay profile based on the phase reference symbol (PRS38. However, the method works in the frequency domain and requires the IFFT with the length of an OFDM symbol period, which results in complex control circuits and high computational work.In this paper, an e

8、fficient symbol timing algorithm for the DAB system is proposed. It accomplishes the symbol timing synchronization by utilizing the good autocorrelation of the phase reference symbol in the time domain. With the appropriate parameters selected, the proposed algorithm can offer an optimal tradeoff be

9、tween the performance and complexity. Compared with the algorithm associated with IFFT, the algorithm proposed has a lower complexity and a nicer flexibility with the system performance unchanged.The rest of this paper is organized as follows. Section 1 describes the symbol synchronization algorithm

10、 associated with IFFT. In Section 2, a new symbol timing algorithm for the DAB system is proposed with its performance analyzed. Section 3 presents the simulation results demonstrating the potential of the proposed algorithm. Finally, we conclude the paper in Section 4.1 Symbol Timing Synchronizatio

11、n in Dab SystemsThe structure of a DAB frame is shown in Fig.1. The synchronization channel consists of a null symbol and a phase reference This work was supported by the NSFC under Grant 60332030 and by the 863 Program under Grant 2006AA01Z270.GUO Yi, LIU Gang, GE Jianhua are all with the National

12、key Lab. of Integrated Service Networks in Xidian Univ.10 西安電子科技大學(xué)學(xué)報(bào)（自然科學(xué)版） 第34卷 symbol （PRS ） that are used for frame, frequency and symbol synchronization. FIC(fast information channel is the second channel that contains multiplex reconfiguration information, service component information, and tra

13、ffic information. The last channel is MSC (main service channel which occupies the major part of the transmission frame and carries all the digital audio service components.Fig 1. Transmission frame structureIn general, to get a precise timing the symbol timing synchronization is divided into coarse

14、 symbol timing synchronization and fine symbol timing synchronization. The coarse symbol timing synchronization roughly estimates the start of a frame. The finesymbol timing synchronization further reduces the residual symbol timing error of the coarse symbol timing synchronization.1.1 Coarse Symbol

15、 Timing SynchronizationThe null symbol is the first symbol in the DAB frame and no signal is transmitted during the null symbol period9, and thus the coarse symbol timing synchronization can be accomplished by null symbol detection through the symbol energy detection5,6. However, the timing-varying

16、property of the signal energy in fading channels makes the energy detection difficult, so the energy ratio of two consecutive blocks should be used instead.1.2 Fine Symbol Timing SynchronizationThe PRS following the null symbol in the DAB frame is a dedicated pilot symbol 9. Thus the CIR h(n is obta

17、ined byh (n =IFFTZ *(k Z r (k , (1where Z (k is the k -th subcarrier of PRS in the frequency domain, Z r (k is of a received PRS, Z *(k is the complex conjugate of Z (k , denotes the element-by-element product of vectors and n is the sample index in the time domain. With the estimated CIR, the fine

18、symbol timing synchronization can be accomplished.The CIR-based symbol synchronization method is effective, but it requires too much computational work. The length of IFFT in (1 is the same as a useful symbol period excluding the guard interval, and thus the computational load of symbol synchronizat

19、ion is identical to that of OFDM demodulation.2 Proposed Efficient Symbol Timing Synchronization2.1 Ideas for Simplification Figure 2 Autocorrelation of the signal z Figure 3 Autocorrelation of part of the signal z (the first 40 points 增 刊 郭 漪等：OFDM 系統(tǒng)中一種有效的符號(hào)定時(shí)同步方案 11The phase reference symbol in t

20、he frequency domain is defined in DAB standard 9, so the phase reference symbol in the time domain can be achieved byz =z (0z (1 z (N 1 , (21z (n =N Z (k exp(j2kn /N , n =0, 1, , N 1,k =0N 1where z (n is the phase reference symbol in the time domain. Fig.2 shows the autocorrelation of the signal z .

21、 It can be seen that the signal z has very good autocorrelation just as PN sequences, and thus the fine symbol timing synchronization can be accomplished by moving the signal z to correlation. Considering that part of the signal z also has good autocorrelation as shown in Fig.3 and the precision of

22、coarse symbol timing can be improved through the energy ratio detection, the correlation can be simplified by only moving part of z in a narrow range. Simulation and measurement results show that the moving correlation only needs to be done about 20 times around the coarse symbol timing position in

23、AWGN channels or 2060 times in Rayleigh channels.2.2 Proposed AlgorithmAssuming that the coarse symbol timing position through energy ratio detection is P , and that the 2D +1 correlations of the first W points of z with the received signal r (n around P arecor(n =r (n +l conj(z (l , n =P D , P D +1

24、, , P +D . (3l =0W 1And thus the exact start of a frame is the position that shows a peak cor(n . The fine symbol timing synchronization algorithm is illustrated in Fig.4.r Figure 4 Structure of the fine symbol timing synchronization algorithmIt is noted that the proposed algorithm can offer an opti

25、mal tradeoff between the channel environment and the hardware resource. For instance, under bad channel environment, W and D should be chosen to be larger if the hardware resource is enough; otherwise they should be less.Table 1 summarizes the computational requirements of the CIR-based algorithm us

26、ing a 2048-point IFFT5 and the correlation algorithm proposed in this paper. The former method requires N +(N /2log2N =2048+(2048/2log2204813312 times complex multiplication and N log 2N = 2048log2204822528 times complex addition. Assuming W =40 and D =30, the latter only requires W (2D +1 =40(230+1

27、 =2440times complex multiplication and (W 1(2D +1 =(401(230+1 =2379times complex addition, below 20% of the computational load required for the former.T able 1 Comparation of the computational requirementsMethod Num. of Complex Multiplication Num. of Complex AdditionCIR-based algorithm N +(N /2log2N

28、 N log 2Nusing a 2048-point IFFTProposed Method W (2D +1 (W -1(2D +112 西安電子科技大學(xué)學(xué)報(bào)（自然科學(xué)版） 第34卷 3 Simulation ResultsIn this section, we demonstrate the performance of our symbol timing synchronization algorithm through the probability of error for detecting the strongest path (PEDS and the system BER

29、performance. The algorithm is investigated using the Monte-Carlo simulation method in both an additive white Gaussian noise (AWGN channel and a multipath fading channel (thechannel impulse response is shown in Fig.5. We assume that the mode 1 OFDMsignaling is used in the DAB system9 for the simulati

30、on. In this scenario, theOFDM system has 2048 subcarriers among which there are 1536 activesubcarriers with a 1 kHz inter-carriers pacing in total. The length of the guardinterval is 504 samples. The simulation results of symbol timing synchronization with W=40, W =60,W =80, W =100 and W =120 are sh

31、own in Fig.6. Fig.6(a shows the results in anA WGN channel with D=10, and Fig.6(b shows the results in a multipath channelwith D = 30. We observe that the symbol timing synchronization algorithmFigure 5. CIR of multipath fading channel proposed performs well both in AWGN channels and in multipath fa

32、ding channels. If W =120, the PEDS can achieve less than 10-4 when the SNR is -5dB in AWGN channels and less than 10-5 when the SNR is 6dB in multipath fading channels. It is noted that the symbol timing synchronization algorithm proposed performs better with the increase of W but its realization co

33、mplexity is higher, and thus an optimal solution can be derived with the tradeoff between the performance and the complexity. (a Probability of error for detecting the strongest (b Probability of error for detecting the strongestpath in an A WGN channel path in a multipath fading channelFigure 6 Per

34、formance of fine symbol timing synchronizationThe bit-error rate (BER performance of an OFDM system using the proposedsymbol timing synchronization algorithm (W =40,D =30 in a multipath fadingchannel is shown in Fig.7. Here, only the raw bit error without coding andinterleaving is measured. The BER

35、performance of an OFDM system using thealgorithm associated with IFFT5 is shown for calibration purposes. We observethat there is no BER degradation despite the complexity is reduced down to below20%.4 Conclusion Figure 7 BER performance in a multipath fadingWe proposed a computationally efficient s

36、ymbol timing synchronization algorithm. It accomplishes the symbol timing synchronization by utilizing the good autocorrelation of the phase reference symbol in the time domain. Compared with the algorithm associated with IFFT, the algorithm proposed has a lower complexity and a better flexibility w

37、ith the system performance unchanged.增 刊 郭 漪等：OFDM 系統(tǒng)中一種有效的符號(hào)定時(shí)同步方案 13R EFERENCES12345 Taura K. A Digital Audio Broadcasting (DAB Receiver J. IEEE Trans on Consumer Elec, 2000, 42(3:322-326, Vinod B R., Srikanth S. A Null Symbol Detection Algorithm for DAB Receivers C/International Conference on Sig

38、nal Processing. Canada: Communications and Networking, 2007: 312 -315. Moezzi-Madani N, Holisaz. Digital Audio Broadcasting System Modeling and Hardware Implementation C/IEEE Asia Pacific Conference on Circuits and Systems. Canada: IEEE, 2006:1814-1817. Moezzi-Madani N, Rohani E. Hardware Considerations for Digital Audio Broadcasting System C/The 17th International Conference on Microelectronics. Cambridge: , 2005: 209-212. ETS 300 401. Radio Broadcasting System: Digital Audio Broadcasting (DAB to Mobile，Portable and Fixed ReceiversS. Lucioles:ETSI, 2006.6789 Steendam H, Moe