第10卷第10期2010年4月科學(xué)技術(shù)與工程Vol 10 No 10 Apr 20101671-1815(2010)10-2480-04Science Technology and Engineering⊙2010 Sci. Tech. Engng基于酉差分空時(shí)碼的信道佔(zhàn)計(jì)性能研究陳雪松韓明宇(大慶石油學(xué)院電氣信息工程學(xué)院,大慶163000摘要介紹MMSE和LS算法的基本原理同時(shí)給出酉空時(shí)碼的編解碼原理,對(duì)基于OFDM的非盲倌道估計(jì)與基于酉差分空時(shí)碼的盲信道估計(jì)進(jìn)行性能的仿真對(duì)比。其中非盲信道估計(jì)釆用MMSE和LS算法,盲信道估計(jì)采用極大似然算法,通過(guò)誤碼率仿裏曲線對(duì)比可以看出采用酉空時(shí)碼的誤比特率在大信噪比的環(huán)境下要明顯好于采用MMSE和LS;而在小信噪比環(huán)境下,酉空時(shí)碼的誤碼率與采用LS的信道估計(jì)的誤碼率基本一致。關(guān)鍵詞信道估計(jì)正交頻分復(fù)用最小均方誤差最小二乘西空時(shí)碼中圖法分類號(hào)TN96;文獻(xiàn)標(biāo)志碼A在無(wú)線通道中,發(fā)射信號(hào)在傳播過(guò)程中往往會(huì)對(duì)比,但并沒(méi)有給出酉空時(shí)碼針對(duì)非理想信道估計(jì)受到各種物體所引起的遮擋、吸收,從而發(fā)生反射、下的性能對(duì)比,本文將在此給予改進(jìn)。折射和衍射,導(dǎo)致多條路徑信號(hào)分量具有不同的傳播時(shí)延、相位和幅度,并附加有噪聲,最終它們的疊1基于導(dǎo)頻的非盲信道信道佔(zhàn)計(jì)加會(huì)使接收信號(hào)相互抵消或增強(qiáng),產(chǎn)生嚴(yán)重的衰落。因此實(shí)時(shí)的進(jìn)行信道估計(jì)并給予均衡就變得基于導(dǎo)頻的信道估計(jì),即在發(fā)送數(shù)據(jù)流中插入十分重要。常用的信道估計(jì)方法有非盲信道估計(jì)導(dǎo)頻符號(hào),在接收端利用這些已知的導(dǎo)頻符號(hào)進(jìn)行法、半盲信道估計(jì)法、盲信道估計(jì)法。其中非盲信信道估計(jì),導(dǎo)頻符號(hào)可以在同一個(gè)OFDM信息符號(hào)道估計(jì)法主要采用基于導(dǎo)頻序列的最小均方誤差子載波頻率軸方向和時(shí)間軸方向上進(jìn)行二維插入,法(MMSE)和最小二乘法(LS),而盲信道估計(jì)法并或分別在頻率或時(shí)間方向上分別進(jìn)行一維插值,然不發(fā)送專門的訓(xùn)練序列或?qū)ьl,它只是利用接收信后利用差值算法對(duì)信道的其他位置進(jìn)行信道號(hào)本身以及發(fā)送信號(hào)的內(nèi)在特點(diǎn)進(jìn)行信道的估計(jì),估計(jì)4即利用信道的統(tǒng)計(jì)信息(諸如循環(huán)平穩(wěn)特性等)進(jìn)行信道估計(jì)。半盲信道估計(jì)介于盲信道估計(jì)和非數(shù)框誰(shuí)并:頻盲信道估計(jì)之間僅利用少量導(dǎo)頻信號(hào)進(jìn)行信道估輸入調(diào)變計(jì),因而成為目前的研究熱點(diǎn)之一。文獻(xiàn)[1]首先落信道引入了西空時(shí)調(diào)制,文獻(xiàn)[2]給出了了酉空時(shí)碼在瑞麗衰落信道下與理想信道下的非盲信道的性能解變↓估調(diào)換|計(jì)010年1月7日收到圖1典型的基于導(dǎo)頻的OFDM系統(tǒng)模型第一作者簡(jiǎn)介:陳雪松(1972-),女蒙古族黑龍江肇州人副教1.1基于最小二乘算法(LS)的頻域信道估計(jì)授、博士生,研究方向:信息安全,語(yǔ)音信息隱藏S信道估計(jì)就是從最小平方意義上得到的信通信作者簡(jiǎn)介:韓明字(1979-),男漢族黑龍江伊春人,研究道估計(jì)方法,基于LS信道估計(jì)的準(zhǔn)則為:生,研究方向:通信網(wǎng)安全理論與技術(shù)。E-mail::jsf-37@163.XPy(1)10期陳雪松,等:基于酉差分空時(shí)碼的信道估計(jì)性能研究2481式(1)中,y是接收導(dǎo)頻信號(hào),X,是發(fā)送導(dǎo)頻信號(hào)。但是在具有快瑞利衰落的信道環(huán)境下進(jìn)行信道估由式(1)可見(jiàn),基于DS算法的信道估計(jì)較為簡(jiǎn)計(jì)有時(shí)變得非常困難,于是, Hughe提出了酉空時(shí)調(diào)單,計(jì)算量較小,易于實(shí)現(xiàn)但該算法并未利用信道制6來(lái)實(shí)現(xiàn)空時(shí)分組碼的編碼方案,也就是西空時(shí)的頻域和時(shí)域相關(guān)性,同時(shí)忽略了噪聲的影響,致編碼,該方案無(wú)需在接受端進(jìn)行信道估計(jì),但卻能使對(duì)信道估計(jì)的準(zhǔn)確度大大降低。獲得高的數(shù)據(jù)傳輸速率,實(shí)現(xiàn)了盲信道估計(jì),因而12基于最小均方誤差法(MMSE)的信道估計(jì)日益受到人們的關(guān)注。針對(duì)LS算法的上述不足,人們提出了MMSE2.1酉空時(shí)碼的編碼算法,該算法通過(guò)對(duì)均方誤差函數(shù)取極小值來(lái)獲取對(duì)于發(fā)射天線數(shù)為N,且調(diào)制星座為A的系信道估計(jì)響應(yīng)對(duì)于子載波間千干擾和高斯白噪聲具統(tǒng),令G為一群LXL西矩陣,其中L≥N有很好的抑制作用。對(duì)于任何G∈G基于MMSE算法準(zhǔn)則如下:GG=GG=I(3)均方誤差函數(shù)(MSE)為:E|H2.wsE(m)假設(shè)存在一個(gè)N×L矩陣D,以致對(duì)群中的任ws(m)2,對(duì)其求偏導(dǎo)并且令其等于零,可得信酉矩陣G來(lái)說(shuō),DG產(chǎn)生了一個(gè)NxL矩陣,這個(gè)道估計(jì)矩陣矩陣的項(xiàng)是信號(hào)星座集A的元素,即Hp wwsr (m)= PR R-l= FQ Ms Fx"Y(2)DG∈Arxt式(2)中,R=EH"}=RmF"x";則矩陣集DC={DGG∈C}Rrr=Elrr=XFrHn FX+oNIni就構(gòu)成空時(shí)群碼,如果矩陣D滿足D"D= LIQMmsE=R[(FX"XF)"0n+RHH]-x(F"X"XF)1。那么可以得到E為數(shù)學(xué)期望,l為NxN的單位矩陣,F為( DG)(DG)"=LIN在這種情況下,式(5)中的群碼就是空時(shí)酉群NXN的矩陣,且F(kn)=”,0≤k≤N-1,碼。例如,當(dāng)N=2,Nn=2,L=2時(shí),令D=Rm為H的協(xié)方差矩陣Rn為H和Y的互協(xié)方差矩101「0G則DG就是陣3為噪聲方差。MMSE算法通過(guò)分析接收信號(hào)的統(tǒng)計(jì)特性來(lái)對(duì)BPS調(diào)制星座A=|+1,-1|上的群碼。信道進(jìn)行估計(jì),誤碼率要明顯低于LS算法,但其計(jì)對(duì)上述空時(shí)酉群碼進(jìn)行差分調(diào)制編碼如表1,算過(guò)程中涉及到大量的復(fù)數(shù)乘法和矩陣求逆,尤其其中為參考信號(hào)。是對(duì)Qsc求逆,導(dǎo)致計(jì)算復(fù)雜度遠(yuǎn)大于LS算法。表1差分調(diào)制編碼近些年,人們提出了一些改進(jìn)方法,比如利用奇異值分解(SVD)來(lái)化簡(jiǎn)Qw矩陣,同樣達(dá)到了令人信息酉調(diào)制81滿意的效果,但這些都是MMSE的特例,需要滿足差分調(diào)制x=1x1=81x2=x1g2x=x283定的前提條件。22空時(shí)西群碼的譯碼”:12酉空時(shí)碼接收機(jī)可以通過(guò)最大似然(ML)差分譯碼算法或Ⅴ viterbi譯碼算法,本文選擇差分譯碼算法,通過(guò)空時(shí)碼是近些年移動(dòng)通信領(lǐng)域的一個(gè)新的研最大似然差分譯碼算法選擇群中最佳的酉矩陣,經(jīng)究方向,由于同時(shí)在時(shí)域和空域中引入編碼,空時(shí)過(guò)逆映射確定相應(yīng)的比特序列。即當(dāng)前和先前接碼對(duì)同信道干擾和碼間干擾都有較好的抑制作用,收信號(hào)矩陣的差分空時(shí)譯碼為2482科學(xué)技術(shù)與工程10卷Marg maxReTrR g,R F式(8)中,R-1為N8xL矩陣表示第t個(gè)發(fā)射分組4結(jié)束語(yǔ)的接收信號(hào),ReIr代表接收矩陣跡的實(shí)部。本文介紹了一種基于酉空時(shí)碼的低復(fù)雜度的3仿真分析盲信道估計(jì),發(fā)射端和接收端都不需要經(jīng)過(guò)復(fù)雜的數(shù)學(xué)計(jì)算,同時(shí)由于不需要插入導(dǎo)頻信號(hào),頻帶利非盲信道估計(jì)仿真采用一發(fā)一收的OFDM系用率獲得極大提高,通過(guò) Matlab仿真得出了在高信統(tǒng)每個(gè)OFDM符號(hào)的子載波數(shù)為128,帶寬為噪比條件下該方案的性能要優(yōu)于文中提到的另外2MHz,循環(huán)前綴(CP)的長(zhǎng)度為16,采樣率為兩種方案。可見(jiàn)對(duì)于既要提高頻帶利用率,又要提6MH,信道采用瑞利慢衰落信道,系統(tǒng)采用BSK商信號(hào)傳輸質(zhì)量的今天基于兩空時(shí)的盲信追怙計(jì)調(diào)制。肓信道估計(jì)仿真采用二發(fā)一收的 OFDM-M會(huì)有很大的發(fā)展空間。MO系統(tǒng),每一幀有120個(gè)信息比特,假設(shè)一幀內(nèi)信道衰落系數(shù)不變,且?guī)c幀相互獨(dú)立。系統(tǒng)仿真誤I Hochwald B M, Sweldens W. Differential unitary space-time modula-碼率曲線如圖2所示。tion. IEEE Trans Communi, 2000: 48(12)2孔昭煜,王玉龍張軍.基于 MATLAB的差分空時(shí)兩群碼的仿真實(shí)現(xiàn)山西電子技術(shù),2006;(4):41-433 Coler S, Ergen M, Puni A, et al. Channel estimation techniques10based on pilotent in OFDM systems. IEEE Trans on Broadcastin,2006;48(3):223-2294 van de Beck J J Edfors 0. Sandell M, et al. On channel estimation inof DM systems. Vehieular Technology Conference, IEEE 45th, Vol-102me2,25-28,1995;(7):8l5-8191012141618OFDM. EURASIP Journal on Wireless Communicationsking,2005;(2):163-174圖2誤碼率曲線比較圖6 Hughes B L. Differential space-time modulation, IEEE Trans InforTheory,2000;46(7):2567-2578仿真結(jié)果表明,在信噪比小于6dB時(shí),采用酉7 Hochwald B, Marzetta T, Richardson T,a. Systematic desigm of U空時(shí)的盲信道估計(jì)的誤碼率與采用LS的信道估計(jì) nitary甲 ace-time code. IEEE Trans Inform Theory,200046(6)的誤碼率非常接近,但與采用MMsE的信道估計(jì)相1962-1973差3dB左右。當(dāng)信噪比大于10dB左右后,采用酉8(澳 Vucetic B,Yum.空時(shí)編碼技術(shù)王曉梅等澤,北京機(jī)械工業(yè)出版社,2004:123-124空時(shí)的盲信道估計(jì)的誤碼率要逐漸低于采用IS或下轉(zhuǎn)第2486頁(yè))MMSE的信道估計(jì)。我們知道信道估計(jì)不需要插入導(dǎo)頻信號(hào),在信噪比小的時(shí)候很容易導(dǎo)致誤碼率升高,這是它的不足之處,但是它的頻帶利用率要明顯高于采用LS和MMSE的非盲信道估計(jì),這在頻帶資源日益緊張的今天是非常重要的。所以當(dāng)信噪比高于8dB的時(shí)候,采用酉空時(shí)盲信道估計(jì)的優(yōu)勢(shì)就非常明顯了。2486科學(xué)技術(shù)與工程10卷Time-frequency Analysis of Speech Signal Based on Improved S TransformSUN Yan, YU Feng-qinCommunication and Control Engineering Institute, Jiangnan University, Wuxi 214122, P R China)Abstract] STFT has fixed resolution and wavelet transform has a partial phase information and time-scale relations are not suitable for an intuitive visual analysis. Time-frequency analysis of speech signal based on an improvedS transform was proposed. Wavelet transform with spline wavelet as the mother wavelet multiplied by the phase factor was the improved S transform, which had multi-resolution and phase factor and was suitable for the intuitive vis-al analysis. Simulation experiments show the time-frequency structure of the improved S transform is more finecompared with STFT and wavelet transformKey words] speech signal time-frequency analysis improved S transform(上接第2482頁(yè))Unitary Differential Space time Codes Based on theChannel Estimation Performance StudyCHEN Xue-song, HAN Ming-yuInstitute of Electrical and Information Engineering, Daqing Petroleum Institute, Daqing 163318, P R China)Abstract] The MMSE and the LS algorithm, are firstly introduced and given the basic principle of unitaryspace-time codes decoding principles, and then OFDM-based non-blind channel estimation and unitary differentialspace-time codes based on blind channel estimation are used to compare the simulation. One non-blind channel es-timation using the MMSE and the ls algorithm, blind channel estimation using maximum likelihood algorithmBER simulation curves can see that use of unitary space-time codes bit error rate environment in large signal tonoise ratio is significantly better than the use of MMSE and the LS, while in low SNR environment, the unitaryspace-time codes using bit error rate and the LS channel estimation error rate basically the sameKey words] channel estimation OFDM minimum mean-square error least squares unitary dif-ferential space-time codes