TWI377808B - Method for fine symbol timing synchronization in ofdm system - Google Patents

Method for fine symbol timing synchronization in ofdm system Download PDF

Info

Publication number
TWI377808B
TWI377808B TW97112041A TW97112041A TWI377808B TW I377808 B TWI377808 B TW I377808B TW 97112041 A TW97112041 A TW 97112041A TW 97112041 A TW97112041 A TW 97112041A TW I377808 B TWI377808 B TW I377808B
Authority
TW
Taiwan
Prior art keywords
channel
path
frequency division
orthogonal frequency
multiplexing system
Prior art date
Application number
TW97112041A
Other languages
Chinese (zh)
Other versions
TW200943787A (en
Original Assignee
Ali Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ali Corp filed Critical Ali Corp
Priority to TW97112041A priority Critical patent/TWI377808B/en
Publication of TW200943787A publication Critical patent/TW200943787A/en
Application granted granted Critical
Publication of TWI377808B publication Critical patent/TWI377808B/en

Links

Landscapes

  • Synchronisation In Digital Transmission Systems (AREA)

Description

1377808 九、發明說明: 【發明所屬之技術領域】 -種正交頻分複㈣、統+精符料序同步方法 是於:多徑通道的麵线中,在快逮傅立葉轉換之後: 灯精付胡步’以利用插人的離散導頻進行通道估計。 【先前技術】 正交頻分複用(0rth0g0nal frequency_divisi〇n multiplexing,0FDM)在數位通訊的領域中應用多載波調 製(Multi-Carrier Modulation),主要是將通道分成若干正 交子通道’將高速資料信號㈣成並行的低速子資料流 私’调製到在每個子通道上進行傳輪,正交信號可以通過 在接收端採用相關技術來分開,這樣可以減少子通道之間 的相互干擾(ICI) ’每個子通道上的信號帶寬小於通道的 相關帶寬,因此可以減少或消除符號間干擾。 OFDM系統的優點是能有效對抗多徑時延擴展 (multi-path time-delay spread )、且頻譜利用率高,但它也 存在對同步誤差敏感的缺點,同步誤差主要包括載波頻率 偏差、取樣時脈(sampling clock)偏差以及符號同步偏差, 其中符號同步偏差可能會造成符號間干擾(ISI)和子載波 間干擾(ICI),給解調系統帶來嚴重的影響。 且此正交頻分複用技術在數位廣播系統中得到廣泛應 用’以DVB-T系統為例,請參閱第一圖所示之dvB-T系 統的調製和解調流程圖。 在發送史而(transmitter ),於頻域(frequency domain ) 中的符元訊號(symbol)輸入後(10),在插入導頻和傳輸 參數訊息(Transmission Parameter Signaling, TPS )( 11)、 加入保護邊帶(頻帶兩端補零)(12)之後,通過反快速傅 立葉轉換(IFFT) ( 13)調製到互相正交的子載波上,再 於輸出訊號前加上循環字首(cyclic prefix )( 14 ),經過數 位類比轉換電路作一轉換(15),運送至發送前端(16), 得到時域(time domain)上的傳輸資料,並於通道(17) 上發送資料。 接著,相應的接收端(receiver)部分(18)將經過通 道(17)的資料先經過類比數位轉換(19),接著,進行下 變頻及抗混疊濾波(20)。接著,由插值器(21)接收經同 步化的採樣訊號,再進行頻偏相位糾正(22)與粗符號同 步(23),之後去掉循環字首(24),進行快速傅立葉轉換 (FFT)解調(25),解調出傳輸參數訊息(26)後,最後 通過通道估計(estimation)與等化(equalize) (27)得到 被調製資料。其中可利用於執行FFT (25)前或FFT (25) 後進行OFDM系統的同步程序,如圖所示之同步手段,包 括载波同步(29)、採樣(Sampling clock)同步(30)與 精符號(Symbol Timing )同步(28 )手段。其中,精符号 (Fine Symbol Timing)同步(28)的作用是在去除循環字 首時能選擇正確的FFT窗口位置。 上述中’為有效解決複雜的多路徑效應,OFDM系統 即利用加入循環字首,也就是把符元的後面資料複製—分 到前端來’作為保護間隔(Guard Interval),這樣可以避開 訊號因多路徑抵達所造成的干擾。 FFT咖d 中’第—圖與第三圖顯示在多徑通道下的 1自口位置,並由戶s —办 目ρ &八中顯不固口位置產生了非同步的情形, ^ °二、疋為了在去除循環字首時選擇正確的FFT窗口 如此多捏通道下,正確的窗口位置是指第一徑的起 二二。弟一圖所不之多徑通道,此例顯示有第一徑201 1 一 02 ’而習知技術以第-徑201的窗Π位置21為 =的附窗口位置,並不考慮第二徑2〇2的附窗口: 且,圖中斜線部份包括輸出訊號前加上的循環字首 203,204。符號同步一般分為兩個階段: 1)精符號同步,在FFT後執行,檢測剩餘的符號同 y偏差,將FFT窗口的起始位置準蜂地鎖在第一 徑上。 、 2)粗符號同步,在FFT前執行,利賴環字首的相 關性判斷付號起始位置,當訊噪比較低時,粗符號 同步的精度也較低。如第三圖顯示之第一徑3〇1 與第二徑302,在多徑衰落的情況下,粗符號同步 會將窗口位置定在最大徑,此例中就是第二徑3〇2 中的窗口位置31為FFT窗口位置,而非第一徑3〇1 的位置,不利於之後的通道估計與等化,並使得接 收性能下降。 再如弟四圖所示’習知技術在接收端透過通道衝擊塑 應找出正確的FFT窗口。由圖可知,訊號輸入一 〇FDM系 統中(401 )’移除各訊號之前在發送端所加上的循環字首 (403 )’再進行快速傅立葉轉換(405)。此例更接著由訊 號中提取離散導頻(scatter pilots) (407),求出通道頻域 應,並設i=傅=轉換(409) ’得到通道衝擊響 的位置或者最二ί擊響應中第—個超過門檻值 口(4U)。且作為付號起始位置去調節FFT的窗 在以道:=_口進行符號同步,克服上述 包括符=魏1_步誤差產生的干擾, 4 1干擾(ISI)和子載波間干擾(ICI)的問題。 【發明内容】 干擾成的符號間 來消除該影響,主要是在通:在= 月况下,此穩定地檢測出正確的符號同步位置。 本發明之qFDm⑽中符號同步方法 孟旦 速傅立葉轉換模組時,先經由通道“ 讀取其中的離散導頻,藉線性插值得到符號内間 ':又、自、離散導頻,以得到較大的時延容許範圍,曾 逋迢頻域響應。 ^ 透過補零,經反快速傅立葉轉換得出通道衝擊響應, 以此得+出徑的數量、各徑的位置和能量以及各徑卩i的延 遲接著由噪聲功率估計模組根據彳查的位置與數量,將 ,徑分別作為第一徑,依此調整FFT窗口位置,並逐次計 算不,窗口位置時所對應的噪聲功率,最後以噪聲功率最 小的,設為正確的第一徑,所對應的窗口位置即為最佳的 FFT窗口,即準確的符號起始位置。1377808 IX. Invention: [Technical field of invention] - Orthogonal frequency division complex (four), system + fine symbol sequence synchronization method is: in the upper line of multipath channel, after fast catching Fourier transform: Fu Hubu' uses the discrete pilots inserted to perform channel estimation. [Prior Art] Orthogonal Frequency Division Multiplexing (0RDM) applies multi-carrier modulation (Multi-Carrier Modulation) in the field of digital communication, mainly to divide the channel into several orthogonal sub-channels. The signal (4) is parallelized into low-speed sub-data streams and modulated to transmit on each sub-channel. The orthogonal signals can be separated by using relevant techniques at the receiving end, which can reduce mutual interference between sub-channels (ICI). 'The signal bandwidth on each subchannel is less than the associated bandwidth of the channel, so intersymbol interference can be reduced or eliminated. The advantage of OFDM system is that it can effectively resist multi-path time-delay spread and high spectrum utilization, but it also has the disadvantage of being sensitive to synchronization error. The synchronization error mainly includes carrier frequency deviation and sampling time. Sampling clock deviation and symbol synchronization deviation, where symbol synchronization deviation may cause inter-symbol interference (ISI) and inter-subcarrier interference (ICI), which has a serious impact on the demodulation system. And this orthogonal frequency division multiplexing technology is widely used in digital broadcasting systems. Taking the DVB-T system as an example, please refer to the modulation and demodulation flowchart of the dvB-T system shown in the first figure. In the transmission history, after the symbol input in the frequency domain (symbol) is input (10), the pilot and transmission parameter information (Transmission Parameter Signaling, TPS) (11) are inserted, and the protection is added. After the sideband (zero-band complementary to the band) (12), it is modulated onto the mutually orthogonal subcarriers by inverse fast Fourier transform (IFFT) (13), and a cyclic prefix is added before the output signal ( 14), through a digital analog conversion circuit for a conversion (15), transported to the transmitting front end (16), to obtain the transmission data in the time domain (time domain), and send data on the channel (17). Next, the corresponding receiver portion (18) first converts the data passing through the channel (17) by analog-to-digital conversion (19), followed by down-conversion and anti-aliasing filtering (20). Then, the synchronized sampling signal is received by the interpolator (21), and the frequency offset phase correction (22) is synchronized with the coarse symbol (23), and then the cyclic prefix (24) is removed to perform a fast Fourier transform (FFT) solution. After adjusting (25), the transmission parameter message (26) is demodulated, and finally the modulated data is obtained by channel estimation and equalization (27). It can be used to perform the synchronization procedure of the OFDM system before or after the FFT (25), as shown in the synchronization means, including carrier synchronization (29), sampling (Sampling clock) synchronization (30) and fine symbols. (Symbol Timing) Synchronous (28) means. Among them, the Fine Symbol Timing synchronization (28) is used to select the correct FFT window position when the cyclic prefix is removed. In order to effectively solve the complex multipath effect, the OFDM system uses the addition of the cyclic prefix, that is, the data after the symbol is copied to the front end to be used as a guard interval (Guard Interval), so as to avoid the signal factor. Interference caused by multi-path arrival. In the FFT coffee d, the first and third graphs show the position of the self-port under the multipath channel, and the non-synchronous situation occurs in the position of the ρ & Second, in order to remove the loop vocabulary when selecting the correct FFT window so many pinch channels, the correct window position refers to the first two. In the case of the multipath channel, the first path 201 1 - 02 ' is displayed in this example, and the conventional technique uses the window position 21 of the first path 201 as the window position of the =, and does not consider the second path 2附2 attached window: Moreover, the slashed part of the figure includes the cyclic prefix 203, 204 added before the output signal. Symbol synchronization is generally divided into two phases: 1) Fine symbol synchronization, executed after FFT, detecting the remaining symbols with the same y deviation, and locking the starting position of the FFT window to the first path. 2) The coarse symbol synchronization is performed before the FFT, and the correlation of the prefix of the Leilai ring determines the starting position of the paying number. When the signal noise is low, the precision of the coarse symbol synchronization is also low. As shown in the third figure, the first path 3〇1 and the second path 302, in the case of multipath fading, the coarse symbol synchronization will set the window position to the maximum diameter, in this case the second path 3〇2 The window position 31 is the FFT window position, not the position of the first path 3〇1, which is disadvantageous for subsequent channel estimation and equalization, and the reception performance is degraded. As shown in the fourth figure, the conventional technique finds the correct FFT window through the channel impact at the receiving end. As can be seen from the figure, the signal is input into a FDM system (401) to perform a fast Fourier transform (405) on the cyclic prefix (403) added to the transmitting end before removing each signal. In this example, the scatter pilots (407) are extracted from the signal, and the channel frequency domain is determined, and i=fu=conversion (409)' is obtained to obtain the channel impact or the second response. The first one exceeds the threshold value (4U). And as the starting position of the paying position to adjust the FFT window in the channel: = _ port for symbol synchronization, to overcome the above interference = Wei 1 _ step error caused by interference, 4 1 interference (ISI) and inter-subcarrier interference (ICI) The problem. SUMMARY OF THE INVENTION The interference is eliminated between the symbols, mainly in the pass: in the case of = month, this stably detects the correct symbol synchronization position. In the qFDm (10) symbol synchronization method of the present invention, the Mengdan speed Fourier transform module first reads the discrete pilots in the channel through the channel, and obtains the inner symbol 'by the linear interpolation by the linear interpolation to obtain a larger The allowable range of delay, Zeng Wei frequency domain response. ^ Through the zero-padding, the inverse impulse Fourier transform is used to obtain the channel impulse response, which gives the number of exit paths, the position and energy of each path, and the path 卩i The delay is then followed by the noise power estimation module according to the position and number of the check, the diameter is taken as the first path, and the FFT window position is adjusted accordingly, and the noise power corresponding to the window position is calculated successively, and finally the noise power is used. The smallest, set to the correct first path, the corresponding window position is the best FFT window, that is, the exact symbol start position.

其中較佳貫施例係包括’開始時,訊號進入此OFDM 1377808 系統,此接收端粗符號同步將經過通道的資料去掉循環字 首,並接著進行快速傅立葉轉換,經提取各符號中的離散 導頻後,再經下述線性插值後,得到符號内間隔較小的離 散導頻,用於計算通道頻域響應,於補零後,執行反快速 傅立葉轉換得到通道衝擊響應。 接著,先設定一門檻值,根據此門檻值在通道衝擊響 應中找到並記錄多徑的位置和數量等資訊,再根據多徑的 位置和個數,嘗試將每條徑作為第一徑放在窗口起始位 置,分別進行相應的FFT窗口調整,之後通過提取離散導 頻、計算通道頻域響應、計算通道衝擊響應等步驟得到通 道衝擊響應,再計算相應每次不同位置FFT窗口所對應的 通道衝擊響應的噪聲功率。 最後經比對,以通道衝擊響應的噪聲功率最小時相對 應的徑為正確的第一徑位置,也就是最佳的符號起始位置。 【貫施方式】 本發明提供一種0FDM系統中精符號時序同步方法,主 要特徵是0FDM系統中,於快速傅立葉轉換(FFT)之後執 行精符號同步,以找到最佳的FFT視窗位置,消除符號間 干擾(ISI),從而實現接收機的最佳接收性能,請先參閱第 五圖所示之DVB-T糸統的離散導頻插入結構。 如圖所示,各導頻訊號(p i 1 〇 t ’以圖中貫心‘圓圈表不) 的插入位置具有一定規律,在頻率方向(橫向排列)來看, 除了開頭(Κπύη)和結尾(K_〇兩個子載波,在一個符號 内,每隔12個子載波(載有通訊資料,以空心圓圈表示) 10 1377808 會插入一個離散導頻’在時間方向(列向排列)來看,每4 個符號為循環週期(如標號1,2,3,4與標號5,6,7,8的導 頻位置一樣)’規律地重複各導頻的位置。由第1個符號到 第4個符號,插入離散導頻的起始位置會依次差3個子载 波,如圖式_導頻501、502、503、504兩兩相距3個子載 波。如此循環,每4個符號’其離散導頻位置會重複一次。 由於離散導頻的資訊對於接收端是已知的,且插入位 置具有相等的間隔’所以可由方程式(1)求出通道頻域變 應: Λ Η ,,Preferably, the method includes: at the beginning, the signal enters the OFDM 1377808 system, and the coarse symbol synchronization at the receiving end removes the cyclic prefix from the data of the channel, and then performs fast Fourier transform, and extracts the discrete guide in each symbol. After the frequency, the following linear interpolation is performed to obtain a discrete pilot with a small interval within the symbol, which is used to calculate the frequency domain response of the channel. After zero-padding, the inverse fast Fourier transform is performed to obtain the channel impulse response. Then, first set a threshold value, find and record the position and quantity of the multipath in the channel impulse response according to the threshold value, and then try to place each path as the first path according to the position and number of multipaths. The starting position of the window is respectively adjusted by the corresponding FFT window, and then the channel impulse response is obtained by extracting the discrete pilot, calculating the frequency domain response of the channel, calculating the channel impulse response, and then calculating the channel corresponding to the FFT window at each different position. The noise power of the impulse response. Finally, the corresponding path is the correct first path position, that is, the optimal symbol start position, when the noise power of the channel impulse response is minimum. [Complete Mode] The present invention provides a fine symbol timing synchronization method in an OFDM system. The main feature is that in the OFDM system, fine symbol synchronization is performed after fast Fourier transform (FFT) to find an optimal FFT window position and eliminate intersymbols. Interference (ISI) to achieve the best receiver performance, please refer to the DVB-T system's discrete pilot insertion structure shown in Figure 5. As shown in the figure, the insertion position of each pilot signal (pi 1 〇t ' with the center of the circle 'the circle' is not shown) has a certain regularity, in the frequency direction (horizontal arrangement), except for the beginning (Κπύη) and the end (K _ 〇 two subcarriers, within one symbol, every 12 subcarriers (carrying communication data, indicated by open circles) 10 1377808 will insert a discrete pilot 'in the time direction (column alignment), every 4 The symbols are cyclic (as the pilot positions of labels 1, 2, 3, 4 and 5, 6, 7, 8) 'repetitively repeat the position of each pilot. From the first symbol to the fourth symbol The starting position of the inserted discrete pilot will be 3 subcarriers in sequence, as shown in the figure _ pilots 501, 502, 503, 504 are separated by 3 subcarriers. Thus, every 4 symbols 'the discrete pilot position will Repeat once. Since the information of the scattered pilot is known to the receiving end and the insertion positions have equal intervals', the channel frequency domain response can be obtained from equation (1): Λ Η ,,

P P Lk —一(1) 其中是=[画+3><(/111〇£14)+12^^ = 0,1,2...义-1, 並且,士α·表示第1個符號第k個子载波上估計出的 ,道頻域響應,&表示經過快速傳立葉轉換後接收到的 第1個符號第k個子載波上的離散導頻資料,表示該 點十,知的離散導頻資料,*表示求共軛(con juggeK 表示每個符號内離散導頻的個數。 pPP Lk - one (1) where = [drawing + 3 >< (/111 〇 £ 14) + 12 ^ ^ = 0, 1, 2 ... meaning -1, and, ± α · indicates the first The estimated frequency domain response on the kth subcarrier of the symbol, & represents the scattered pilot data on the kth subcarrier of the first symbol received after the fast Fourier transform, indicating the point ten, the known discrete Pilot data, * indicates conjugate (con juggeK indicates the number of discrete pilots within each symbol. p

接著將通道頻域響應之後補零,形成N/2點數據,N r 錢讀循财首後的子魏(Sub-贿i er)個數 .'系統令,2K模式時,N等於_,8K模式時,Ν 號i麻1、92’ Μ附視窗長度,也即移除循環字首後的符 用補零是為了將計算點數凑成2的冪次方,以可使 ::^立葉變換來計算。之後,再經過反快速傅立葉轉 換了求出通道衝擊響應: 11 1377808 hi,n =IFFT[^Lk) 5 n=;l, 2,......, N/2-1 ——---------------------------------(2) 方程式(2)顯示的通道衝擊響應反映出時域通道的多 徑資訊,包括徑數、各徑的位置和能量以及通道最大時延 (channel delay)。 如第六圖所示之多徑通道下的通道衝擊響應示意圖, 其中顯示進入0FDM系統中徑的數量與各徑間的延遲狀Then the channel frequency domain response is followed by zero padding to form N/2 point data, and N r money reads the number of sub-bribes after the first fiscal position. 'System order, in 2K mode, N equals _, In the 8K mode, the nickname i1, 92' Μ is attached to the window length, that is, after the cycle prefix is removed, the zero padding is used to make the calculation point a power of 2, so that::^ The lobe transformation is used to calculate. After that, the channel impulse response is obtained by inverse fast Fourier transform: 11 1377808 hi, n =IFFT[^Lk) 5 n=;l, 2,..., N/2-1 ——-- -------------------------------(2) The channel impulse response shown by equation (2) reflects the number of time domain channels. Path information, including the number of paths, the position and energy of each path, and the channel delay. A schematic diagram of the channel impulse response of the multipath channel as shown in the sixth figure, which shows the number of the diameters entering the 0FDM system and the delay between the paths.

態,除了徑以外,其他為噪聲,如接近能量為零位置的部 份。 上述離散導頻插入結構中,各個符號内離散導頻的位 置間隔是12個子載波’由它們計算得到的通道頻域響應相 當於對真實的通道頻域響應進行了 1/12的取樣,反映^眭 蜮衝擊響應上則是真實衝擊響應的1/12。如果最大通道延 遲超過Tu/12 (Tu為一個符號的週期),則會產生混疊。在State, except for the path, is noise, such as a part close to zero energy. In the above discrete pilot insertion structure, the position interval of the discrete pilots in each symbol is 12 subcarriers. The channel frequency domain response calculated by them is equivalent to sampling the real channel frequency domain response by 1/12, reflecting ^ The 眭蜮 shock response is 1/12 of the true shock response. If the maximum channel delay exceeds Tu/12 (Tu is a symbol period), aliasing occurs. in

SPN (slngiefrequenCynetw〇rk)網路中,這樣的最 延谷♦乾圍是不夠的0 Γ 句】奴南敢大時延的容許範圍,必須通過時間方ρ 的插值以減少離料狀_間隔。方法通常有三種 1) 直接法:直接將連續四個#號的離散導頻提取< 成個付號間隔為3的離散導頻。 2) 2插值法:利用前後複數個(如7個)符號ή C行線性插值得到—個間隔較小(如3: 圖月頻,線性插值的具體過程如下,請參閱》 弟七圖中縱方向表示符號1,橫方向表示每個符费 12 2 M+3 尸 -〜-(3) 、~~(4) ^4,4+3^ =Γ^2,4+3ρ +i~Y6 Υ^=4Υ^Ρ+^Υ1η3ρ 舉例如圖’方程式(3)顯示分別透過標 —-2、k=4)與標號7〇4位置(卜6、 ^位置 ::標請位置(W、k=4 =據計 中央,故係數使用1/2;方程式點的 ,算出標號驗置(1=4、k=7)的插值據 位置並非該兩點的中.央,而县古片A 而‘波7〇2的 數3/4與1/4。藉標號7〇1 * 7{^二=插,=同的係 以減少離散導頻之間的間隔,並且,:上計算據 於其他相對位置的計算。 v驟將應用. 3)另有方法是利用前後多個(通常>7)符號的離 =,。經過FIR濾波得到—個符號間隔為3的離散導 上述三種插值方法都可以使最大時延的容In the SPN (slngiefrequenCynetw〇rk) network, such a maximum delay is not enough. 奴 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 】 There are usually three methods: 1) Direct method: directly extract the four consecutive ## discrete pilots into a discrete pilot with a payout interval of 3. 2) 2 interpolation method: using a plurality of (such as 7) symbols ή C line linear interpolation before and after - a small interval (such as 3: graph monthly frequency, the specific process of linear interpolation is as follows, please see) The direction indicates the symbol 1, and the horizontal direction indicates that each fee is 12 2 M+3 corpse-~-(3), ~~(4) ^4,4+3^ =Γ^2,4+3ρ +i~Y6 Υ ^=4Υ^Ρ+^Υ1η3ρ For example, the equation (3) shows the position through the mark -2, k=4) and the number 7〇4 respectively (Bu 6, ^ position:: mark position (W, k= 4 = According to the center, the coefficient is 1/2; the equation point is calculated. The interpolation position of the label verification (1=4, k=7) is not the middle of the two points, and the county ancient film A and ' The number of waves 7〇2 is 3/4 and 1/4. By the number 7〇1 * 7{^2=plug, = the same system to reduce the interval between discrete pilots, and: the upper calculation is based on other relatives The calculation of the position will be applied. 3) Another method is to use the multiple of the (usually >7) symbols before and after. After FIR filtering, a discrete guide with a symbol interval of 3 can be used. Make the maximum delay

Tu/3,, 力也依次遞增,而本發明之較佳實_係應 = 少離散導頻之間的間隔。 插值減 經上述計算第七圖所示的運算,得到間隔為3的離散 1377808 導頻,通過公式(1)、(2)計算得到通道衝擊響應,將通 過設定門檻值來判斷有效的徑(path)的位置和個數。 2 mean _cir=—hln ( 5) ^ n=\ path _ threshold = kx mean _ cir (6) 其中,i為通道衝擊響應,長度為N/2。為 通道衝擊響應的幅度的均值。6為比例因數,可以根據實 際應用設置,根據通道衝擊響應的統計特性設置,例如在 DVB-T系統中,免可以設置為5。 通過公式(5)、(6),門檻值可以與通道衝擊響應的均 值成比例,且超過門捏值的即為有效的徑,否則就是雜訊。 精符號同步完成後,在通道衝擊響應中的延遲最大的那根 徑對應的時延即為通道最大時延(Channel Delay Spread),因此本發明所提供的演算方法除了可以得到正確 的窗口位置,.還能獲得最大時延資訊和噪聲功率。 ’如習知技術所述,在多徑通道下,粗符號同步會將符 號起始位置鎖在最大徑位置,當第一徑能量小於第二徑. 時,窗口位置會定在第二徑附近。此時,如果直接根據求 得的通道衝擊響應結果,以第一個過門檻值的徑作為第一 徑,將會出現誤判。如第八圖所示,當第一徑能量小於第 二徑時,即如圖中顯示經過精符號同步後的通道衝擊響應。 如第六圖與第八圖所示,通道衝擊響應中除了徑以外 的點都是雜訊,如接近零的低能量部分,雜訊的功率可以 表示為方程式(7): W = WISI + WAWGN + W1CI --------------------(7) 14 13*77808 其中灰/57為符號間干擾引入的雜訊’ 為通道的南 斯雜訊,%σ為子載波間干擾引入的雜訊。 在時不變通道(time-invariant channel)中,灰洲似, 基本保持不變,如果FFT窗口位置找錯,如第一徑在窗口 以外,則灰m和灰/C/都會增加,導致總的噪聲功率妒變大。 本發明即提出根據正確的位置其噪聲功率最小這一判斷准 則,通過比較每種可能的FFT窗口位置的噪聲功率,得到 最佳的窗口位置。 而在時變通道(time-variant channel)..下,由.於通. 道的時間選擇性衰落(Time-Selective Fading)特性的影 響,連續符號的%_隨時間會有一定的波動,即使FFT 窗口不變,不同符號得到的噪聲功率也有所不同,在不同 符號内調節窗口位置計算相應的噪聲功率,可能出現正確 位置的噪聲功率比錯誤位置的大,從而引起誤判。因此, 本發明進一步提出在計算噪聲功率前,可以先預存一定數 量(大於一個符號)的時域資料(FFT前),之後每次在此 預存的時域資料中移動視窗,只從中獲取一個符號的資料 輸出到快速傅立葉轉換模組,確保精符號同步模組每次計 算的噪聲功率都取自同一個符號,消除在時間上波動 的影響。為簡化計算,噪聲功率W的值可以由通道衝擊響 應的絕對值求和來得出。 本發明所提供利用噪聲功率判定正確的第一徑的過程 具體如第九圖所示之流程,包括先將上述流程得出的通道 衝擊響應,透過設定的門檻值,可得出複數個徑(步驟 S901);接著,將找到的第一條徑往左移至通道衝擊響應的 15 1377808 起始位置,如第六圖或是第八圖通道衝擊響應圖式的左 方,本發明則以延遲FFT窗口的方式實現左移的動作,如 將時域上的FFT窗口起始位置延遲相應的資料個數(步驟 S903),將新窗口位置的FFT輸出通過提取間隔為12的離 散導頻,計算通道頻域響應(步驟S905),補零到N/2,然 後經過IFFT得到通道衝擊響應(步驟S907),通過求IFFT 輸出的絕對值和,得到第一次噪聲功率結果(步驟S909 )。 之後,依次將找到的其餘徑左移到衝擊響應的起始位 置,同樣是FFT窗口起始位置延遲相應的資料個數,計算 其餘的噪聲功率結果。 本發明所提出的方法係不需移動所有的複數個徑,只 要經比對程序後判斷第一次最佳徑的位置(步驟S911), 如後一次的噪聲功率比前一次大,左移過程即可停止,說 明前一次的窗口位置是該左移過程中最佳的。如果穆到最 後一個徑(如第Μ條)而未出現比前一次噪聲功率大的情 況,則說明第Μ條徑才是左移過程中的最佳位置。 經上述將徑左移的過程結束後,則開始右移過程,從 最後一徑開始依次右移至衝擊響應的起始位置,要實現該 過程只需將時域上的FFT窗口起始位置提前相應的資料個 數(步驟S913)。 將提前的新窗口位置的FF7輸出通過提取的離散導 頻,並接著計算通道頻域響應(步驟S915),經補零與經 過IFFT得到通道衝擊響應(步驟S917),通過求IFFT輸 出的絕對值和,得到(估計)噪聲功率(步驟S919)。 同樣,經比對得到右移方向最佳徑的位置(步驟 16 1377808 S921 )’而比對結束的條件同樣是後一次的噪聲功率比前一 次大,或者移到第一徑(此例係由最後一徑開始^ ^ 才出現噪聲功率最小值。 〇^ 隶後,比較左移與右移兩個方向的最小嗓聲功去,較 小的那個才是本發明使用的正確的第一徑和移動方二= 驟 S923)。 ^Tu/3,, the force is also incremented, and the preferred embodiment of the invention should = the interval between the less discrete pilots. The interpolation is reduced by the operation shown in the seventh figure above to obtain the discrete 1377808 pilot with interval 3. The channel impulse response is calculated by formulas (1) and (2), and the effective path is determined by setting the threshold value. The location and number of ). 2 mean _cir=—hln ( 5) ^ n=\ path _ threshold = kx mean _ cir (6) where i is the channel impulse response and the length is N/2. The mean of the magnitude of the channel impulse response. 6 is the scaling factor, which can be set according to the actual application settings and according to the statistical characteristics of the channel impulse response. For example, in the DVB-T system, it can be set to 5. Through the formulas (5) and (6), the threshold value can be proportional to the mean value of the channel impulse response, and the effective value is exceeded when the threshold value is exceeded. Otherwise, it is noise. After the fine symbol synchronization is completed, the delay corresponding to the path with the largest delay in the channel impulse response is the channel delay spread, so the calculation method provided by the present invention can obtain the correct window position. Can also get the maximum delay information and noise power. 'As described in the prior art, under the multipath channel, the coarse symbol synchronization will lock the symbol start position at the maximum path position. When the first path energy is smaller than the second path, the window position will be set near the second path. . At this time, if the path of the first threshold is used as the first path based on the obtained channel impulse response result, a misjudgment will occur. As shown in the eighth figure, when the first radial energy is smaller than the second diameter, the channel impulse response after the fine symbol synchronization is displayed as shown in the figure. As shown in the sixth and eighth diagrams, the points other than the path in the channel impulse response are all noise, such as the low-energy part near zero, and the power of the noise can be expressed as equation (7): W = WISI + WAWGN + W1CI --------------------(7) 14 13*77808 where ash/57 is the noise introduced by intersymbol interference' as the channel's Nans noise, %σ is the noise introduced by inter-subcarrier interference. In the time-invariant channel, the gray-scale is similar and remains basically unchanged. If the FFT window position is wrong, if the first path is outside the window, both gray m and gray/C/ will increase, resulting in total The noise power becomes larger. The present invention proposes a criterion for determining the noise power of the correct position based on the correct position, and obtains the optimum window position by comparing the noise power of each possible FFT window position. Under the time-variant channel, the %_ of continuous symbols will fluctuate with time, even if it is affected by the time-selective Fading characteristic of the channel. The FFT window is unchanged, and the noise power obtained by different symbols is also different. The corresponding noise power is calculated by adjusting the window position in different symbols, and the noise power at the correct position may be larger than the error position, thereby causing misjudgment. Therefore, the present invention further proposes that before calculating the noise power, a certain number (more than one symbol) of the time domain data (before the FFT) can be pre-stored, and then each time in the pre-stored time domain data, the window is moved, and only one symbol is obtained therefrom. The data is output to the fast Fourier transform module, ensuring that the noise power calculated by the fine symbol synchronization module is taken from the same symbol each time, eliminating the influence of fluctuations in time. To simplify the calculation, the value of the noise power W can be derived by summing the absolute values of the channel impulse response. The process for determining the correct first path by using the noise power is specifically as shown in the ninth figure, and the channel impulse response obtained by the above process is firstly passed through the set threshold value to obtain a plurality of paths ( Step S901); Next, moving the first path found to the left to the initial position of the channel impulse response 15 1377808, as shown in the left or right side of the channel impulse response pattern of the sixth or eighth figure, the present invention delays The FFT window realizes the left shifting action, such as delaying the start position of the FFT window in the time domain by the corresponding number of data (step S903), and calculating the FFT output of the new window position by extracting the discrete pilot with the interval of 12, and calculating The channel frequency domain response (step S905) is zero-padded to N/2, and then the channel impulse response is obtained by IFFT (step S907), and the first noise power result is obtained by finding the sum of the absolute values of the IFFT outputs (step S909). After that, the remaining paths found are moved to the left of the impact response in turn, and the corresponding number of data is delayed by the start position of the FFT window, and the remaining noise power results are calculated. The method proposed by the present invention does not need to move all the multiple paths, as long as the position of the first best path is judged after the comparison procedure (step S911), if the next noise power is larger than the previous one, the left shift process You can stop, indicating that the previous window position is the best during the left shift. If Mu goes to the last path (such as the first line) and does not appear to be larger than the previous noise power, then the third path is the best position during the left shift. After the process of shifting the left direction to the left is completed, the right shifting process is started, and the rightward shift from the last path to the start position of the impact response, and the process needs to advance the start position of the FFT window in the time domain. The corresponding number of pieces of data (step S913). The FF7 output of the advanced new window position is passed through the extracted discrete pilot, and then the channel frequency domain response is calculated (step S915), and the channel impulse response is obtained by zero padding and IFFT (step S917), and the absolute value of the IFFT output is obtained. And, the (estimated) noise power is obtained (step S919). Similarly, the position of the best path in the right shift direction is obtained (step 16 1377808 S921 )' and the condition of the end of the comparison is also that the next noise power is larger than the previous time, or moved to the first path (this example is At the beginning of the last path ^ ^, the minimum noise power appears. 〇^ After the sequel, compare the minimum squeaking work in the left and right directions, the smaller one is the correct first path and used in the present invention. Move the second party = step S923). ^

再請參閱第十圖,其顯示本發明使用之精符號同步渖 异法的系統示意圖,與習知技術(如第一圖)不同的是, 本發明先利用線性插值得到間隔為3或其他複數值的=散 導頻,經計算通道頻域響應與補零後,再經過IFFT得到^ 道衝擊響應,經此同步過程調整FFT窗口位置,以估計噪 聲功率,並藉比較噪聲功率判斷第一徑的正確 到最佳的m窗口起始位置。 ^ 圖中所不之系統包括有接收訊號後進行移除循環字首 的模、’且101由於發送端在傳送訊號時,會因為要避開前Referring again to the tenth figure, which shows a system diagram of the fine symbol synchronization different method used in the present invention, unlike the prior art (such as the first figure), the present invention first uses linear interpolation to obtain an interval of 3 or other complex numbers. The value of the = scattered pilot frequency, after calculating the frequency domain response and zero padding, and then the IFFT to obtain the channel impulse response, the FFT window position is adjusted by the synchronization process to estimate the noise power, and the first path is judged by comparing the noise power. Correct to the best m window starting position. ^ The system in the figure does not include the module to remove the cyclic prefix after receiving the signal, and 101 because the transmitting end is transmitting the signal, it will be avoided.

一個符元因多路徑延遲抵達所造成的訊號干擾,即於輸出 訊遗财加上彳純字首’並於此接收端需將其移除後,才能 進行其他動作.。 、,接著讯號通過快速傅立葉轉換模組102,經轉換後傳 =出去,然而,為避免多路徑效應產生的訊號干擾,需要 得到正確的FFT自〇進行符號同纟,故輸出的訊號將經由 通道頻域響應計算模組103先提取出離散導頻, 精由上述 線性插值得到符號内間隔較小(如3)的離散導頻,以得 到較大的時縣城®,再湘上述方程式(1)計算通道 頻域響應。 (S ) 17 1377808 之後,符號内先補零,以將計算點數湊成2的冪次方 (如DVB-T系統中,計算點數為N/2,N為FFT窗口長度, 也即移除循環字首後的符號長度),再經反快速傅立葉轉換 模組104,由通道衝擊響應計算模組106計算通道衝擊響 應,以此得出徑的數量、各徑的位置和能量以及各徑間的 延遲。 接著,通過門檻值的各徑訊息傳送至噪聲功率估計模 組105,此模組105根據徑的位置與數量,並將各徑作為 第一徑並對應FFT窗口,每次都調整FFT窗口位置,再經 上述模組101,102, 103, 104等模組根據通道衝擊響應計算 模組106產生的各徑數據,逐筆計算所對應的噪聲功率, 最後,以所計算的噪聲功率最小的徑設為正確的第一徑, 以此能得出最佳的FFT窗口,即準確的符號起始位置。 起始位置找尋完畢後,再次通過提取離散導頻、IFf;T 轉換、得出通道衝擊響應,根據門檻值找到最後一徑的位 置,得到通道最大時延,並根據門檻值以下各點的能量計 算平均噪聲功率。 上述找尋FFT窗口的過程中,除了在連續的符號中進 行噪聲功率估計,亦可以透過另一儲存單元107儲存同一 個符號中的數據,以在同一個符號中估計噪聲功率,消除 訊號在時間上的波動對噪聲功率判斷的影響。 根據上述精符號同步系統的各模組之運作整理出如第 十一圖所示的符號同步方法之較佳實施例之流程。 於步驟Sill中,訊號進入此OFDM系統,此接收端將 經過通道的資料去掉循環字首,並接著進行快速傅立葉轉 18 1377808 換(步驟S113),之後,經上述線性插值後,得到符號内 間隔較小的離散導頻(步驟S115),並再提取各符號中的 離散導頻(步驟S117),在較佳實施例中,係先在FF7後 提取複數個(如7個)離散導頻,進行線性插值得到一個 符號内間隔為較小(如3)的離散導頻,以提高最大時延 的容許範圍。 接著,於步驟S119計算通道頻域響應,較佳實施例係 以利用方程式(1)計算通道頻域響應,經通道頻域響應補 零後,.執行反快速傅立葉轉換得到通道衝擊響應(步驟 S121 )。根據此門檻值在通道衝擊響應中找到並記錄多徑的 位置和數量等資訊(步驟S123)。 根據上述多徑的位置和個數,嘗試將每條徑作為第一 徑放在窗口起始位置,分別進行相應的FFT窗口調整(步 驟S125),調整過程可以在連續的符號中進行,也可以儲 存一定資料在同一個符號中進行。再從每次窗口調節後的 FFT結果中提取一個符號的離散導頻,如同之前間隔為12 的離散導頻,再計算相應每次FFT窗口的噪聲功率(步驟 S127)。 之後重複步驟117, 119, 121, 123, 125與127等步驟, 包括重複提取離散導頻、計算通道頻域響應、執行IFFT 計算通道衝擊響應等步驟(1Π、119、121),於得出並記 錄徑的資訊之後(步驟S123),再進行FFT窗口調整(步 驟S125),再計算相應每次FFT窗口的噪聲功率(步驟 S127)。 經反覆上述步驟後,最後根據噪聲功率的比對結果, 19 1377808 能得到正確的第一徑位置,也就是噪聲功率最小的最佳的 符號起始位置(步驟S129)。 綜上所述,本發明為一 0FDM系統中精符號時序同步方 法,主要是利用線性插值得到間隔較小的離散導頻,經計 ' 算通道頻域響應、補零後,經過IFFT得到通道衝擊響應, • 並根據最小雜訊功率推測進行判斷第一徑的步驟,藉以得 到正確的FFT窗口起始位置,也就是經此同步的步驟得到 0FDM系統中最佳的符號起始位置。 • 惟以上所述僅為本發明之較佳可行實施例,非因此即 侷限本發明之專利範圍,故舉凡運用本發明說明書及圖示 内容所為之等效結構變化,均同理包含於本發明之範圍 内,合予陳明。 【圖式簡單說明】 第一圖所示為習知技術之DVB-T系統的調製和解調電路 不意圖, φ 第二圖所示為多徑通道下的FFT窗口位置示意圖之一; 第三圖所示為多徑通道下的FFT窗口位置示意圖之二; 第四圖所示為習知技術於OFDM系統中執行符號同步的 電路示意圖·, • 第五圖所示為DVB-T系統的離散導頻插入結構; 第六圖所示為多徑通道下的通道衝擊響應示意圖; 第七圖所示為内插離散導頻的插入結構示意圖; 第八圖所示為多徑通道下的通道衝擊響應示意圖; 20 1377808 第九圖係為本發明利用噪聲功率判定第一徑的流程; 第十圖所示為本發明使用之精符號同步演算法的系統示意 圆, 第十一圖顯示本發明0FDM系統中符號同步方法之較佳實 施例之流程。A signal interferes with the arrival of a symbol due to multipath delay, that is, after the output of the message is added to the pure prefix, and the receiver needs to remove it, other actions can be performed. Then, the signal is transmitted through the fast Fourier transform module 102, and then converted to go out. However, in order to avoid the signal interference caused by the multipath effect, it is necessary to obtain the correct FFT auto-symbol symbol, so the output signal will be The channel frequency domain response calculation module 103 first extracts the scattered pilots, and obtains the discrete pilots with small inter-symbol intervals (such as 3) by the above linear interpolation to obtain a larger time county®, and then the above equation (1) ) Calculate the channel frequency domain response. After (S) 17 1377808, the symbol is first padded to make the calculated number of points into a power of 2 (as in the DVB-T system, the number of points is N/2, N is the length of the FFT window, that is, the shift In addition to the symbol length after the beginning of the cycle prefix, the channel impulse response calculation module 106 calculates the channel impulse response through the inverse fast Fourier transform module 104, thereby obtaining the number of paths, the position and energy of each path, and the paths. The delay between. Then, the path information of the threshold value is transmitted to the noise power estimation module 105. The module 105 adjusts the FFT window position according to the position and number of the diameters, and takes each path as the first path and corresponds to the FFT window. Then, the modules 101, 102, 103, 104 and the like calculate the corresponding noise power according to the path data generated by the channel impulse response calculation module 106, and finally, the path with the minimum calculated noise power is set. For the correct first path, the best FFT window is obtained, which is the exact starting position of the symbol. After the starting position is found, the discrete pilot, IFf;T conversion is extracted again, and the channel impulse response is obtained. The position of the last path is found according to the threshold value, and the maximum delay of the channel is obtained, and the energy of each point below the threshold value is obtained. Calculate the average noise power. In the process of searching for the FFT window, in addition to performing noise power estimation in consecutive symbols, the data in the same symbol may be stored through another storage unit 107 to estimate the noise power in the same symbol, and the signal is eliminated in time. The effect of fluctuations on noise power judgment. According to the operation of each module of the fine symbol synchronization system described above, the flow of the preferred embodiment of the symbol synchronization method shown in Fig. 11 is organized. In step Sill, the signal enters the OFDM system, and the receiving end removes the cyclic prefix from the data of the channel, and then performs fast Fourier transform 18 1377808 (step S113), after which the inter-symbol interval is obtained after the linear interpolation. a smaller discrete pilot (step S115), and extracting discrete pilots in each symbol (step S117). In the preferred embodiment, a plurality of (eg, 7) discrete pilots are first extracted after FF7. Linear interpolation is performed to obtain a discrete pilot with a small inter-symbol interval (such as 3) to increase the allowable range of maximum delay. Next, in step S119, the channel frequency domain response is calculated. In the preferred embodiment, the channel frequency domain response is calculated by using equation (1), and after the channel frequency domain response is zero-padded, the inverse fast Fourier transform is performed to obtain the channel impulse response (step S121). ). Information such as the position and number of multipaths is found and recorded in the channel impulse response based on this threshold value (step S123). According to the position and the number of the multipaths, try to place each path as the first path at the start position of the window, and perform corresponding FFT window adjustments respectively (step S125), and the adjustment process may be performed in consecutive symbols, or may be performed. Store certain data in the same symbol. Then, a discrete pilot of one symbol is extracted from each window-adjusted FFT result, like the discrete pilot with a previous interval of 12, and the noise power of each corresponding FFT window is calculated (step S127). Then, steps 117, 119, 121, 123, 125, and 127 are repeated, including steps of repeatedly extracting discrete pilots, calculating a channel frequency domain response, performing an IFFT calculation channel impulse response, and the like (1, 119, 121). After the information of the track is recorded (step S123), the FFT window adjustment is performed (step S125), and the noise power of each corresponding FFT window is calculated (step S127). After repeating the above steps, finally, according to the comparison result of the noise power, 19 1377808 can obtain the correct first path position, that is, the optimum symbol start position with the smallest noise power (step S129). In summary, the present invention is a fine symbol timing synchronization method in an OFDM system, which mainly uses linear interpolation to obtain a discrete pilot with a small interval. After calculating the frequency domain response of the channel, after zero-padding, the channel impact is obtained through IFFT. In response, • and the step of determining the first path based on the minimum noise power estimation, thereby obtaining the correct starting position of the FFT window, that is, the step of synchronizing obtains the optimal symbol starting position in the 0FDM system. The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Therefore, equivalent structural changes in the description and the drawings of the present invention are equally included in the present invention. Within the scope of the agreement, Chen Ming. [Simple description of the diagram] The first figure shows the modulation and demodulation circuit of the DVB-T system of the prior art. φ The second figure shows one of the FFT window positions under the multipath channel. The second schematic diagram of the FFT window position under the multipath channel is shown; the fourth figure shows the circuit diagram of the conventional technique for performing symbol synchronization in the OFDM system. · The fifth figure shows the discrete guide of the DVB-T system. The frequency insertion structure; the sixth figure shows the channel impulse response under the multipath channel; the seventh figure shows the insertion structure of the interpolated discrete pilot; the eighth figure shows the channel impulse response under the multipath channel. 20 1377808 The ninth diagram is the flow of determining the first path using noise power according to the present invention; the tenth figure shows the system schematic circle of the fine symbol synchronous algorithm used in the present invention, and the eleventh figure shows the 0FDM system of the present invention; The flow of a preferred embodiment of the medium symbol synchronization method.

【主要元件符號說明】 訊號輸入1〇 加入保護邊帶12 加上循環字首14 去掉循環字首24 通道估計與等化27 數位類比轉換15 接收前端18 下變頻及抗混疊濾波 頻偏相位糾正22 傳輸參數訊息26 採樣同步30 第一徑201 窗口位置21 第一徑301 窗口位置31 接收訊號401 快速傅立葉轉換405 插入導頻和TPS 11 IFFT 13 通道17 FFT 25 載波同步29 發送前端16 類比數位轉換19 20 插值器21 粗符號同步23 精符号同步28 第二徑202 循環字首203,204 第二徑302 移除循環字首403 提取離散導頻407 21 1377808 反快速傅立葉轉換409 找FFT窗口 411 導頻 5(U、502、503、504 標號 701,702, 703, 704, 705, 706 移除循環字首模組101 快速傅立葉轉換模組102 通道頻域響應計算模組103 反快速傅立葉轉換模組104 噪聲功率估計模組105 衝擊響應計算模組106 儲存單元107[Main component symbol description] Signal input 1〇Add protection sideband 12 plus loop word first 14 Remove loop prefix 24 Channel estimation and equalization 27 Digital analog conversion 15 Receive front end 18 Down conversion and anti-aliasing filter frequency offset phase correction 22 Transmission Parameter Message 26 Sampling Synchronization 30 First Path 201 Window Position 21 First Path 301 Window Position 31 Receive Signal 401 Fast Fourier Transform 405 Insert Pilot and TPS 11 IFFT 13 Channel 17 FFT 25 Carrier Synchronization 29 Transmit Front End 16 Analog Digit Conversion 19 20 Interpolator 21 coarse symbol synchronization 23 fine symbol synchronization 28 second path 202 cyclic prefix 203, 204 second path 302 removal of cyclic prefix 403 extraction of discrete pilot 407 21 1377808 inverse fast Fourier transform 409 find FFT window 411 pilot 5 (U, 502, 503, 504, 701, 702, 703, 704, 705, 706 removal cycle prefix module 101 fast Fourier transform module 102 channel frequency domain response calculation module 103 anti-fast Fourier transform module 104 noise Power estimation module 105 impulse response calculation module 106 storage unit 107

22twenty two

Claims (1)

1377808 十、申請專利範圍: 1. 一種正交頻分複用系統中精符號時序同步方法,包括 、有: 接收訊號; 進行一快速傅立葉轉換; 執行一線性插值,以得到間隔較小的離散導頻; 提取線性插值後的離散導頻; 由提取之離散導頻計算通道頻域響應; 執行一反快速傅立葉轉換,計算通道衝擊響應與一有 效之徑的門檻值,將通道衝擊響應與該門檻值比 較,得到複數個徑的資訊; 根據該複數個徑的資訊,重複調整符號的FFT窗口位 pp · 旦 , 提取離散導頻、計算通道頻域響應與計算通道衝擊響 應等步驟,得到一個以上相對於該徑的通道衝擊響 應,並計算噪聲功率;以及 根據該一個以上的噪聲功率的比對結果,得到該符號 之正霉起始位置。 2. 如申請專利範圍第1項所述之正交頻分複用系統中精 符號時序同步方法,其中該線性插值係利用該符號前 後複數個符號的離散導頻進行該線性插值得到一個符 號間隔較小的離散導頻。 3. 如申請專利範圍第1項所述之正交頻分複用系統中精 符號時序同步方法,其中於計算該通道頻域響應後, 23 1377808 該符號補零,以形成具有2的冪次方的資料個數。 4.如申請專利範圍第1項所述之正交頻分複用系統中精 符號時序同步方法,其中經該通道衝擊響應計算後, 反映出該正交頻分複用系統之時域通道的多徑資訊, 包括徑數、各徑的位置和能量以及該通道最大時延。 ' 5.如申請專利範圍第4項所述之正交頻分複用系統中精 符號時序同步方法,其中將各徑作為該正交頻分複用 系統之第一徑,設定在該FFT窗口起始位置,並調整 φ 相應的FFT窗口,以計算相對之噪聲功率.。. 6. 如申請專利範圍第5項所述之正交頻分複用系統中精 符號時序同步方法,其中以相對該噪聲功率為最小的 徑決定該FFT窗口的位置。 7. 如申請專利範圍第1項所述之正交頻分複用系統中精 符號時序同步方法,其中該FFT窗口調整過程在連續 的符號中進行。 8. 如申請專利範圍第1項所述之正交頻分複用系統中精 φ 符號時序同步方法,其中該FFT窗口調整過程在.同一 符號中進行。 9. 如申請專利範圍第8項所述之正交頻分複用系統中精 • 符號時序同步方法,其中係先預存一大於一個符號的 . 時域資料,之後每次移動該FFT窗口都只是從中獲取 一個符號的資料,以確保每次計算的噪聲功率的% 部分都取自同一個符號。 10. 如申請專利範圍第1項所述之正交頻分複用系統中精 符號時序同步方法,其中該噪聲功率係由該通道衝擊 24 1377808 響應的絕對值求和求出。 11.如申請專利範圍第1項所述之正交頻分複用系統中精 符號時序同步方法,其中該符號之起始位置相對於該 正交頻分複用系統之第一徑,該第一徑之判斷步驟包 括有: • 設定一門檻值,由該通道衝擊響應得出複數個徑; 將該複數個徑依次往左移至該通道衝擊響應的起始位 置; # 將相對該徑之FFT窗口位置的FFT輪出通過提取灕散 導頻; 計算相對該徑之通道頻域響應; 補零到具有2的奉次方的資料個數, 經過該反快速傅立葉轉換得到相對該徑之通道衝擊響 應, 計算相對該徑之噪聲功率; 經比對相對於該複數個往左移的徑的噪聲功率,判斷 • 最佳徑的位置; 將該複數個徑依次往右移至該通道衝擊響應的起始位 置; ' 將相對該徑之FFT窗口位置的FFT輸出通過提取離散 導頻; 計算相對該徑之通道頻域響應; 補零到具有2的冪次方的資料個數; 經過該反快速傅立葉轉換得到相對該徑之通道衝擊響 25 應; =對讀徑之噪聲功率; 、相對於該複數個往 佳經的位置;以及 寿夕的杈的噪聲功率,判斷最 比較往左移與往右移兩 m 疋該正交頻分複用系乾向^取小噪聲功率,以決 12·如申請專利範圍第^項弟一控和移動方向。 符號時序同步方法,龙、^之士正交頻分複用系統中精 位置延遷相應的資料個數; 3.如申請專利範圍第n項 …二二左移的動·作。... 符號時序同步方法,、地之正交頻分複用系統中精 位且提前相應的資 ^的FFT窗π起始 14·如申請專觸 ‘現㈣心移的動作。 符號時序同步方法,以分複用系統中精 響應的絕對值求和求出㈣率係由該通道衝擊 15. -種正交頻分複用系統 序 :於一多徑通道的正交頻分複用系統二 二口的判斷得到該系統中符號的 透. 驟包括有: 见置’该方法步 接收訊號; 將經過通道的訊號去掉循環字首; 進行一快速傅立葉轉換; 執行-線性插值’以得到間隔較小的離散導頻 提取線性播值後的離散導頻; 、 26 於 ,頻计""通這頻域響應; 方的資料個數; 補令以形成具有2的冪次 執轉換,計算通道衝擊響應與-有 較,得到複數^的H逼衝擊響應與該門捏值比 根據該複數個徑的資訊 提取離散導頻、計#通道窗口; 應等步驟,得:一二:f:應與計算通道衝擊響 應,並計㈣聲功率;=對於該#的通道衝擊響 比了上的噪聲功率,以相對該噪聲功率為最 =禮決定該FFT窗口的位置,得到該符號之起始 位置。 16·= j利乾圍第15項所述之正交頻分複用系統中精 付號時序同步方法,其中經該通道衝擊響應計算後, 反映出該正交頻分複用系統之時域通道的多徑 包括讀、各經的位置和能量以及該通道最大時延。 17. 如申請翻範圍㈣項所述之正交頻分複㈣統中精 符號時f同步方法,其中將各徑作為該正交頻分複用 系統之第-徑’設定在該FF丁窗口起始位置,並調整 相應的FFT窗口,以計算相對之噪聲功率。 18. 如申料利範㈣15項所述之正交齡複^统中精 符號時序同步方法,其中該FFT窗口調整過程在連續 的符號中進行。 19.如申請專利範圍第15項所述之正㈣分複n统中精 27 13.77808 符號時序同步方法,其中該FFT窗口調整過程在同一 符號中進行。 20. 如申請專利範圍第19項所述之正交頻分複用系統中精 符號時序同步方法,其中係先預存一大於一個符號的 ' 時域資料,之後每次移動該FFT窗口都只是從中獲取 - 一個符號的資料,以確保每次計算的噪聲功率都取自 同一個符號。 21. 如申請專利範圍第15項所述之正交頻分複用系統中精 • 符號時序同步方法,其中該噪聲功率係由該通道衝擊 響應的絕對值求和求出。 C S ) 281377808 X. Patent application scope: 1. A fine symbol timing synchronization method in an orthogonal frequency division multiplexing system, comprising: receiving signals; performing a fast Fourier transform; performing a linear interpolation to obtain a discrete guide with a smaller interval Frequency; extracting the linearly interpolated discrete pilot; calculating the frequency domain response of the channel from the extracted discrete pilot; performing an inverse fast Fourier transform to calculate the threshold response of the channel and an effective path threshold, and the channel impulse response and the threshold Comparing the values, obtaining the information of the plurality of paths; according to the information of the plurality of paths, repeating the steps of adjusting the FFT window of the symbol pp · den, extracting the scattered pilot, calculating the frequency domain response of the channel, and calculating the channel impulse response, etc., obtaining one or more The channel impact response is calculated relative to the path, and the noise power is calculated; and based on the comparison of the one or more noise powers, the starting position of the symbol is obtained. 2. The fine symbol timing synchronization method in an orthogonal frequency division multiplexing system according to claim 1, wherein the linear interpolation performs the linear interpolation by using a discrete pilot of the plurality of symbols before and after the symbol to obtain a symbol interval. Smaller discrete pilots. 3. The fine symbol timing synchronization method in the orthogonal frequency division multiplexing system according to claim 1, wherein after calculating the frequency domain response of the channel, 23 1377808 the symbol is zero-padded to form a power of 2 The number of data of the party. 4. The fine symbol timing synchronization method in the orthogonal frequency division multiplexing system according to claim 1, wherein the time domain channel of the orthogonal frequency division multiplexing system is reflected by the impulse response calculation of the channel Multipath information, including the number of paths, the position and energy of each path, and the maximum delay of the channel. 5. The fine symbol timing synchronization method in the orthogonal frequency division multiplexing system according to claim 4, wherein each path is set as the first path of the orthogonal frequency division multiplexing system, and is set in the FFT window. Start position and adjust the corresponding FFT window of φ to calculate the relative noise power. 6. The fine symbol timing synchronization method in an orthogonal frequency division multiplexing system according to claim 5, wherein the position of the FFT window is determined by a path that is the smallest relative to the noise power. 7. The fine symbol timing synchronization method in an orthogonal frequency division multiplexing system according to claim 1, wherein the FFT window adjustment process is performed in consecutive symbols. 8. The fine φ symbol timing synchronization method in the orthogonal frequency division multiplexing system according to claim 1, wherein the FFT window adjustment process is performed in the same symbol. 9. The fine symbol timing synchronization method in the orthogonal frequency division multiplexing system described in claim 8 wherein the time domain data is stored for more than one symbol, and then the FFT window is moved each time. Obtain a symbol from it to ensure that the % portion of the noise power calculated each time is taken from the same symbol. 10. The fine symbol timing synchronization method in the orthogonal frequency division multiplexing system according to claim 1, wherein the noise power is obtained by summing the absolute values of the response of the channel impact 24 1377808. 11. The fine symbol timing synchronization method in an orthogonal frequency division multiplexing system according to claim 1, wherein a starting position of the symbol is relative to a first path of the orthogonal frequency division multiplexing system, the first The determining step of the path includes: • setting a threshold value, and the plurality of paths are obtained from the impact response of the channel; moving the plurality of paths to the left to the beginning of the impact response of the channel; # will be relative to the path The FFT wheel position of the FFT window is extracted by extracting the divergent pilot; calculating the frequency domain response of the channel relative to the path; zero-padding to the number of data having a power of 2, and obtaining the channel relative to the path through the inverse fast Fourier transform Impulse response, calculating the noise power relative to the path; determining the position of the optimal path by comparing the noise power with respect to the plurality of paths moving to the left; moving the plurality of paths to the right to the right The starting position of the FFT output relative to the FFT window position of the path by extracting the scattered pilot; calculating the frequency domain response of the channel relative to the path; zero-padding to the number of data having a power of 2 After the inverse fast Fourier transform, the channel impact response 25 to the path is obtained; = the noise power of the read path; the position relative to the plurality of good passages; and the noise power of the 夕 杈, the judgment is the most Shift left and shift to the right by two m 疋 The orthogonal frequency division multiplexing system takes the small noise power to determine the control direction and the moving direction. Symbol timing synchronization method, the number of data corresponding to the precise positional delay in the Orthogonal Frequency Division Multiplexing system of Dragon and ^; 3. If the patent application scope is the nth item... ... symbol timing synchronization method, the FFT window π start of the precision bit in the Orthogonal Frequency Division Multiplexing system and the corresponding resource in advance. 14 If the application is specifically touched, the action of the current (four) heart shift. The symbol timing synchronization method is obtained by summing the absolute values of the fine response in the sub-multiplexing system. (4) The rate is impacted by the channel. 15. Orthogonal frequency division multiplexing system sequence: orthogonal frequency division in a multipath channel The judgment of the second or second port of the multiplexing system results in the transmission of the symbols in the system, including: see the method of receiving the signal; removing the cyclic prefix from the signal passing through the channel; performing a fast Fourier transform; performing - linear interpolation Extracting the linear pilot value of the discrete pilot with a small interval of discrete pilots; 26, the frequency meter "" the frequency domain response; the number of data; the order to form a power of 2 Performing the conversion, calculating the channel impulse response and comparing with the H-impact response of the complex number ^ and extracting the discrete pilot, the #channel window according to the information of the plurality of paths; Two: f: should calculate the channel impulse response, and count (four) sound power; = for the # channel impact ratio of the upper noise power, relative to the noise power is the most = the position of the FFT window, get the Symbolic Starting position. 16·= j. The time-series synchronization method of the fine-paying number in the orthogonal frequency division multiplexing system described in Item 15, wherein the time domain of the orthogonal frequency division multiplexing system is reflected after the impulse response calculation of the channel The multipath of the channel includes the reading, the position and energy of each passage, and the maximum delay of the channel. 17. If the application is to reverse the orthogonal frequency division (4) in the range (4), the f-synchronization method is used, wherein each path is set as the first-path ' of the orthogonal frequency division multiplexing system in the FF window. Start position and adjust the corresponding FFT window to calculate the relative noise power. 18. The method of synchronizing the fine symbol timing in the orthogonal age complex as described in Item 15 of the application (4), wherein the FFT window adjustment process is performed in consecutive symbols. 19. The method of claim 4, wherein the FFT window adjustment process is performed in the same symbol as described in claim 15 of the patent application. 20. The fine symbol timing synchronization method in the orthogonal frequency division multiplexing system according to claim 19, wherein the time domain data greater than one symbol is prestored, and then the FFT window is only moved from each time. Get - A symbolic material to ensure that the noise power calculated each time is taken from the same symbol. 21. The fine symbol timing synchronization method in an Orthogonal Frequency Division Multiplexing system according to claim 15, wherein the noise power is obtained by summing the absolute values of the channel impulse responses. C S ) 28
TW97112041A 2008-04-02 2008-04-02 Method for fine symbol timing synchronization in ofdm system TWI377808B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW97112041A TWI377808B (en) 2008-04-02 2008-04-02 Method for fine symbol timing synchronization in ofdm system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW97112041A TWI377808B (en) 2008-04-02 2008-04-02 Method for fine symbol timing synchronization in ofdm system

Publications (2)

Publication Number Publication Date
TW200943787A TW200943787A (en) 2009-10-16
TWI377808B true TWI377808B (en) 2012-11-21

Family

ID=44869099

Family Applications (1)

Application Number Title Priority Date Filing Date
TW97112041A TWI377808B (en) 2008-04-02 2008-04-02 Method for fine symbol timing synchronization in ofdm system

Country Status (1)

Country Link
TW (1) TWI377808B (en)

Also Published As

Publication number Publication date
TW200943787A (en) 2009-10-16

Similar Documents

Publication Publication Date Title
US7394876B2 (en) Enhanced channel estimator, method of enhanced channel estimating and an OFDM receiver employing the same
US10218459B2 (en) Apparatus and methods for symbol timing error detection, tracking and correction
JP4159030B2 (en) Timing synchronization method for wireless networks using OFDM
CN101686216B (en) Fine symbol timing synchronization method and device for orthogonal frequency division multiplexing system
JP5144687B2 (en) OFDM Code Timing Recovery Method and System
CN102195906A (en) Pilot signal designing method and system for orthogonal frequency division multiplexing (OFDM) system
TW200934189A (en) Methods and apparatus for estimating the channel impulse response
Li et al. Super-imposed training scheme for timing and frequency synchronization in OFDM systems
CN101552758A (en) Accurate symbol sequential synchronous method in orthogonal frequency division multiplexing (OFDM) system
CN101895309B (en) A kind of frequency offset regulation method based on signal-to-noise ratio (SNR) estimation
CN101729479A (en) Blind channel estimation method based on cyclostationarity of OFDM signals
CN109412987A (en) A kind of ofdm system channel tracking method
TWI300657B (en)
James et al. Channel estimation for OFDM systems
Mousa et al. Channels estimation in OFDM system over Rician fading channel based on comb-type pilots arrangement
TWI377808B (en) Method for fine symbol timing synchronization in ofdm system
WO2011066776A1 (en) Method and apparatus for correcting sample frequency difference in ofdm system
KR101275852B1 (en) Method and apparatus for transmitting/receiving data based on sc-fde using uw
TWI429248B (en) Intersymbol interference removal method and receiver
WO2014127862A1 (en) Ofdm packets time synchronisation
EP1387544A2 (en) Synchronisation in multicarrier receivers
KR101276036B1 (en) Transmitter for modulating using cyclic delay diversity and receiver for estimating residual timiming offset of ofdm system
JP5925137B2 (en) Digital broadcast receiving apparatus and digital broadcast receiving method
CN102571646A (en) Timing estimation and adjustment method and device in OFDM (Orthogonal Frequency Division Multiplexing) system
CN101119349A (en) Improved OFDM signal modulating, demodulating method

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees