基于迭代算法的SFBC-OFDM系统

SFBC-MIMO系统碰撞信号恢复算法
胡琦;杨峰;丁良辉;钱良
【期刊名称】《信息技术》
【年(卷),期】2015(39)12
【摘要】为了提升无线传输速率,空频编码(SFBC)以及多输入多输出-正交频分复用(MI-MO-OFDM)技术被引入到无线局域网(WLAN)标准中,而在无线局域网中,由隐藏终端导致的信号碰撞问题会极大影响WLAN系统的通信性能.针对上述问题,提出了一种基于SFBC MI-MO-OFDM系统的碰撞信号恢复算法(SCD-MIMO),该算法通过碰撞信号重构、接收信号求解和空频解码三个步骤对碰撞信号进行恢复.从仿真结果可以看出在信噪比为12时系统误码率仅为10-4,证明文中提出的算法能够有效解决SFBC MIMO-OFDM系统的信号碰撞问题.
【总页数】5页(P42-45,50)
【作者】胡琦;杨峰;丁良辉;钱良
【作者单位】上海交通大学电子工程系,上海200240;上海交通大学电子工程系,上海200240;上海交通大学电子工程系,上海200240;上海交通大学电子工程系,上海200240
【正文语种】中文
【中图分类】TN919.3
【相关文献】
1.基于二进制搜索算法的RFID系统防碰撞算法 [J], 邓洁;程良伦
2.基于后退式二进制搜索算法的有源RFID系统防碰撞算法 [J], 王静;盛磊
3.RFID系统防碰撞算法比较分析及其改进算法 [J], 吴跃前;辜大光;范振粤;杜明辉
4.基于二进制搜索算法的RFID系统防碰撞算法 [J], 孙文胜;马建波
5.倒频谱分析在碰撞声发射源信号恢复中的研究 [J], 张国强;闫勇;胡永辉
因版权原因，仅展示原文概要，查看原文内容请购买。

TD-LTE 培训考试1. LTE物理层采用带有循环前缀的正交频分多址(OFDMA)技术作为下行多址方式，采用具有单载波特性的单载波频分多址(SC-FDMA)技术作为上行多址方式。

2. E-UTRA的L1是按照资源块(RB)的方式来使用频率资源的，以适应可变的频谱分配。

一个资源块在频域上包含12个宽度为15kHz的子载波。

3. LTE采用扁平化网络结构，E-UTRAN主要由eNodeB构成。

4. LTE小区平均吞吐量反映了一定网络负荷和用户分布情况下的基站承载效率，是网络规划重要的容量评价指标。

5. 与下行OFDM不同，上行SC-FDMA在任一调度周期中，一个用户分得的子载波必须是_连续的。

6. LTE支持Hard handover only切换方式。

7. 在同样的覆盖要求下，采用F频段组网与采用D频段组网相比，所需要的站点数更少。

8. 为什么用符号末端部分复制为循环前缀：保证时域信号连续9. 哪个步骤可以把多个OFDM子载波转换成单信号传输：IFFT10. 在MIMO模式，哪个因素对数据流量影响最大：发射天线数目11. 哪个信道用来指示PDCCH所用的符号数目：PCFICH12. 支持LTE的UE的最大带宽是：20 MHz13. 在OFDM中，子载波间隔F和符号时间T的关系是：f = 1/t14. 1.4MHz的带宽中，一个子帧中用于承载PDSCH的资源约占：1/215. 哪种RLC模式可以使业务时延最小：Transparent Mode (TM)16. 传送主同步信号和辅同步信号需要多大带宽：1.08 MHz17. 以下哪些带宽是TDD-LTE支持的：20 MHz、10 MHz、5 MHz、1.4 MHz18. 在LTE中，上行链路降低峰均比(RAPR)的好处是：增强上行覆盖、降低均衡器复杂度、降低UE功率损耗19. LTE规划过程中，影响小区覆盖半径的因素有：系统带宽、传播模型、天线模式、小区边缘规划速率20. 路测时发现小区间天线接反可以从那几个部分去排查：核查小区PCI参数是否配错、排查BBU-RRU光纤是否接反、排查小区间RRU-天线间的跳线是否接反21. 在系统消息上查看LTE终端能力时，从NPO的角度，主要需关注UE 的那些方面能力和特性：支持的频段、支持的加密算法、支持的传输模式、支持的终端能力等级、是否支持同频异频切换22. LTE的物理层上行采用SC-FDMA 技术，下行采用OFDMA技术23. PDSCH信道的TM3模式在信道质量好的时候为开环空分复用，信道质量差的时候回落到单流波束赋型24. LTE要求下行速率达到100Mbps ，上行速率达到50Mbps;UE的切换方式采用硬切换。

SFBC-OFDM系统在快衰落信道下的性能分析
叶霞;林云
【期刊名称】《信息技术》
【年(卷),期】2006(30)11
【摘要】在STBC-OFDM系统的基础上介绍了SFBC-OFDM系统的原理、编译码方案,并通过仿真做出了SFBC-OFDM系统在快衰落信道下的性能,同时与STBC-OFDM系统的性能进行比较.结果表明:SFBC-OFDM系统在室内环境即时间延迟缓慢的情况下性能与STBC-OFDM基本上达到一致,在室外环境快衰落情况下性能要比STBC-OFDM系统要好.
【总页数】4页(P44-47)
【作者】叶霞;林云
【作者单位】重庆邮电学院,重庆,400065;重庆邮电学院,重庆,400065
【正文语种】中文
【中图分类】TN911.22
【相关文献】
1.快衰落信道条件下无线中继系统波束形成方案 [J], 樊娜;张健;张驭龙;姚清
2.MIMO系统在空时二维相关莱斯快衰落信道下的性能 [J], 赖国庭;尹俊勋;喻华文;林凡
3.规则LDPC码在瑞利快衰落信道下的译码算法性能分析 [J], 解福洲
4.快衰落信道下改进的STBC解码性能分析 [J], 杨柳;刘汉奎;郝希准;李骏
5.快时变衰落信道下OFDM系统的SAGE检测算法研究 [J], 狄碎虎;卓嘎;魏润国;胡东升;姜军;董志诚;;;;;;
因版权原因，仅展示原文概要，查看原文内容请购买。

基于迭代算法的STBC-OFDM系统
聂圣峰;安翠珍;庞伟正;裴颖
【期刊名称】《应用科技》
【年(卷),期】2005(032)008
【摘要】为了消除循环前缀对空时分组编码的正交频分复用系统(STBC-OFDM)带宽损失的影响,使用了迭代算法来校正接收到的信号.与传统的带有循环前缀的STBC-OFDM系统相比,带宽效率得到了很大的提高.仿真结果显示,该方法在迭代次数很小时性能便可接近带有循环前缀的STBC-OFDM系统.
【总页数】2页(P10-11)
【作者】聂圣峰;安翠珍;庞伟正;裴颖
【作者单位】哈尔滨工程大学,信息与通信工程学院,黑龙江,哈尔滨,150001;哈尔滨工程大学,信息与通信工程学院,黑龙江,哈尔滨,150001;哈尔滨工程大学,信息与通信工程学院,黑龙江,哈尔滨,150001;哈尔滨工程大学,信息与通信工程学院,黑龙江,哈尔滨,150001
【正文语种】中文
【中图分类】TN914.4
【相关文献】
1.改进的迭代STBC-OFDM系统信道估计与解码算法 [J], 郑勇;冯大政
2.大规模MIMO系统中基于牛顿迭代和超松弛迭代的WWSE预编码算法 [J], 孙文胜; 许俊杰
3.基于反馈迭代算法的航空通信系统信道估计算法研究 [J], 张建康; 赵悠悠; 尚应博; 穆晓敏
4.基于扩展路径识别算法的水声OFDM系统低复杂度迭代稀疏信道估计 [J], 赵世铎;鄢社锋
5.基于分群迭代算法的中药颗粒配药系统药罐动态预编组方法 [J], 张硕瑞;许东来因版权原因，仅展示原文概要，查看原文内容请购买。

填空题1．eNB之间通过X2接口通信,进行小区间优化的无线资源管理。

2．E-UTRAN系统在1.4MHz、3MHz、5MHz、10MHz、15MHz和20MHz带宽中，分别可以使用___6个、15个、25个、50个、75个和100个RB。

3．LTE系统由于采用了OFDM技术，因此来自用户之间的干扰很小，主要干扰是小区间干扰。

4．OFDM符号中的__GI____可以克服符号间干扰。

5．PDSCH信道的调制方式有QPSK、16QAM_和64QAM。

6．从整体上来说，LTE系统架构仍然分为两个部分，即EPC和_E-UTRAN___。

7．对于LTE物理层的多址方案，在下行方向上采用基于CP的OFDMA，在上行方向上采用基于CP的_SC-FDMA___。

8．一个RB由若干个RE组成，频域宽度为_180___kHz，时间长度为0.5ms9．LTE中，下行的峰值速率为100Mbps ，上行的峰值速率为50Mbps。

单选题1．LTE上下行传输使用的最小资源单位叫做( B)。

A、PRBB、REC、REGD、CCE2．LTE系统中，一个无线帧时间长度为( D)。

A、0.5msB、1msC、5msD、10ms3．LTE协议中所能支持的最大RB个数为（D）个。

A、6B、20C、50D、1104．SC-FDMA与OFDM相比，以下正确的是（D）。

A、能够提高频谱效率B、能够简化系统实现C、没区别D、能够降低峰均比5．多普勒效应引起的附加频移称为多普勒频移，若移动台向远离基站方向移动，则此时因多普勒频移会造成移动台接收频率（A）A、偏小B、偏大C、不变D、以上均有可能6．根据协议对LTE系统需求支持的定义，从空闲态到激活态的时延和零负载（单用户、单数据流）、小IP分组条件下单向时延分别小于多少（B）A、50ms和10msB、100ms和5msC、200ms和5msD、100ms和50ms7．下列对于LTE系统中下行参考信号目的描述错误的是（D）A、下行信道质量测量（又称为信道探测）B、下行信道估计，用于UE端的相干检测和解调C、小区搜索D、时间和频率同步8．下列哪项技术的快速发展和引入使得FDMA技术能够应用到LTE系统中（A）。

Journal of Southeast University (English Edition) Vol.21,No.2,pp.127131June 2005 ISSN 1003—7985Iterative receivers for SFBC-OFDM systemswith adaptive training schemeJing YaXu XiaodongChen MingCheng Shixin(National Mobile Communications Research Laboratory,Southeast University,Nanjing 210096,China)Abstract:This paper considers the design of iterative receivers for space-freguency block-coded orthogonal freguency division multiplexing (SFBC-OFDM)systems in unknown wireless dispersive fading channels.An iterative joint channel estimation and symbol detection algorithm is derived.In the algorithm,the channel estimator performs alternately in two modes.During the training mode,the channel state information (CSI)is obtained by a discrete-Fourier-transform-based channel estimator and the noise variance and covariance matrix of the channel response is estimated by the proposed method.In the data transmission mode,the CSI and transmitted data is obtained iteratively.In order to suppress the error propagation caused by a random error in identifying symbols,a simple error propagation detection criterion is proposed and an adaptive training scheme is applied to suppress the error propagation.Both theoretical analysis and simulation results show that this algorithm gives better bit-error-rate performance and saves the overhead of OFDM systems.Key words:space-freguency block-coded orthogonal freguency division multiplexing (SFBC-OFDM);iterative receiver;channel estimation;adaptive training schemeReceived 2004-08-24.Foundation item:The National High Technology Research and Devel-opment Program of China (863Program)(No.2002AA 123031).The combined application of multi-input and multi-output (MIMO)antenna technology and orthog-onal freguency division multiplexing (OFDM)modula-tion yields improvements to communication capacity and guality to meet the reguirements of the next gener-ation of wireless systems.Recently,several studies ad-dressing the design and application of MIMO-OFDM have been conducted [1-3].In Ref.[4],it is shown that applying the simple orthogonal space freguency coding technigue first proposed in Ref.[5]for OFDM systems can have better performance in fast fading environ-ments and other implementation advantages.In this paper,we focus on the design of iterative receivers for space-freguency block-coded orthogonal freguency division multiplexing (SFBC-OFDM)sys-tems in unknown wireless dispersive fading channels with outer channel coding.The iterative joint channel estimation and symbol detection algorithm is derived.The initial channel estimation is achieved by using the training seguence [6],which not only simplifies the ini-tial channel estimation,but also attains the best estima-tion performance.By this training seguence we also present a method to estimate the noise variance and co-variance matrix of the channel response,which are nee-ded by the subseguent iterative algorithm.In order tosuppress the error propagation caused by a random er-ror in identifying symbols,an adaptive training scheme is applied in the system.For this improved scheme,when a decision-error is estimated by a simple criteri-on,one more training seguence is reguired to be sent by the transmitter to update the initial channel estima-tion.The simulation results show that the improved al-gorithm can prevent error propagation effectively and conseguently improve the bit-error-rate (BER )per-formance and keep a low system overhead.In this paper,(·)T ,(·)H and (·)*denote the transposition,Hermitian transposition and conjugate,re-spectively;if !is a given !"1vector,diag(!)is the !"!diagonal matrix with the entries of !on the main diagonal;and "!is the !"!identity matrix.1 System ModelConsider an SFBC-OFDM system with two trans-mission antennas and one reception antenna.As shown in Fig.1,the information bits are encoded by an outer-channel-code encoder and then interleaved.The inter-leaved code bits are modulated by an M-phase shift ke-ying (MPSK)modulator.Finally,the modulated MPSK symbols are encoded by an SFBC encoder and trans-mitted from two transmission antennas.For simplifica-tion,we assume that a rate 1/2convolutional code is applied as the error correcting code.To minimize the latency of the system,coding is assumed to be per-formed over a single OFDM symbol.The interleaver isFig.1 Transmitter and receiver structure with two transmission antennas and one reception antenna We denote N as the total number of subcarriers,orthe FFT size.Xiis the SFBC symbol vector transmittedfrom the i-th antenna.The time domain channel im-pulse response between the i-th transmission antennaand the reception antenna can be modeled as an L-order tapped-delay-line given by hi ,hi={hi(0),.,h i (l),.,hi(L-1)}T which is a complex Gaussianvector,representing different path gains.The elementsof hi are uncorrelated with zero-mean and variation!2i={!20,.,!2l,.,!2L-1}T.Without loss of generality,we assume the total average power of the channel im-pulse response to be unity,i.e.2L-1l=0!2l=1,so the cor-relation matrix of hiis a diagonal matrix with the ele-ments of!2i on the main diagonal.hiare assumed to beindependent and identically distributed among different antenna couples.At the receiver side,with the assumptions that the guard interval duration is longer than the channel maxi-mum excess delay and that the channel is guasi-station-ary(i.e.the channel does not change within one OFDM symbol duration),then the demodulated signal at the receiving antenna can be represented byY=22i=1diag(Xi)"i+z(1)where z is a white complex Gaussian noise vector withcovariance matrix!2z IN,Hiis the channel freguencyresponse between the i-th transmission antenna and the reception antenna,which can be written in matrix nota-tion as follows:Hi =wNhi(2)where wNis the N XL Fourier transform matrix.Assume the channel freguency responses betweenadjacent subcarriers are nearly constant,i.e.Hi(n)*Hi(n+1),n=0,2,.,N-2.Then the channel fre-guency response vector Hican be divided into the even and odd component vectors.The corresponding ele-ments of two vectors are egual.Each component vector can be achieved by performing an N/2-point discreteGi=w N2hi(3)where w N2is theN2XL Fourier transform matrix madeof the odd rows of wN.According to the operation of the space-freguency encoder described in Ref.[4],the symbol vector X isdivided into the even component vector Xeand the oddcomponent vector Xo.Similarly,Xi,e,Xi,odenote theeven and odd component vectors of Xitransmitted from the i-th transmission antenna.Then the output of the space-freguency encoder can be expressed in terms of the even and odd component vectors asX1,e=Xe,X1,o=-X*o,X2,e=Xo,X2,o=X*e(4)The demodulated signal at the receiving antenna and the noise vector Z can also be rewritten in terms ofthe even and odd component vectors as Y=[Y TeY To]Tand Z=[Z TeZ To]T.Considering the assumption about channel freguency response between adjacent subcarri-ers and the SFBC coding constraints of Eg.(4),Eg.(1)can be further expressed asY=Swh+z(5) withS=diag(Xe)diag(Xo)diag(-X*o)diag(X*e[])w=w N20w N[]2,h=h1h[]2(6)2Iterative Joint Decoding and Channel Esti-mationIn this section,we consider the iterative receiver design for SFBC-OFDM systems in fast fading chan-nels as depicted in Fig.1.In our algorithm,the channel estimator performs alternately in two modes.During the training period,the CSI is obtained by a DFT-based channel estimator assisted by the optimum training se-guence[6].When the systems are in the data transmis-sion mode,the CSI and transmitted data is obtained it-eratively.It consists of an ML SFBC decoder and a821Jing Ya,Xu Xiaodong,Chen Ming,and ChengShixin2.1 Iterative SFBC decoderEg.(5)represents the overall input /output rela-tion of the system as a function of the channel impulse response.According to the maximum likelihood criteri-on,the joint optimal solution of X and h can be ob-tained by(^X !^h )=arg max X !h{p (Y X !h )}(7)The solution of (7)can be achieved by an itera-tion algorithm.This algorithm ensures attainment of atleast a local maximum.Theoretically,it maintains non-decreasing likelihood at each iteration step [7].Given the initial CSI ^h [0],the symbol vector ^X can be obtained conveniently according to Alamouti*s com-bining scheme [5].The estimate of X during the (i +1)-th iteration can be expressed as^X [i +1]e ^X [i +1][]o=diag W N 2^h [i ]()1()*diag W N 2^h [i ]()2diag W N 2^h [i ]()2()*diag -W N 2^h [i ]()[]1Y e Y *[]o(8)After that,the estimated symbol vector ^X[i +1]is deinterleaved and decoded by a hard guantization ML outer-channel-code decoder.Instead of being regarded as the detection results,the detected data is fed back to reconstruct the estimate of S in Eg.(6)and then is used to estimate the channel attenuation again.Ac-cording to the Bayesian philosophy [7],the conditionalmean ^h [i ]obtained during the i -th iteration is ^h [i ]=1!2Z!W H (^S [i ])H Y (9)with!=R -1h +1!2ZW H (^S [i ])H (^S [i ])()W-1(10)where R h and !2Z denote the covariance matrix of h and noise variance,respectively,which can be meas-ured with the aid of training seguence.Due to the orthogonality property of SFBC as well as the constant amplitude modulation,W H (^S [i ])H (^S [i ])W results to N I 2L regardless of the transmitted symbol.(Noting that the average energy of the symbols transmitted from each antenna is nor-malized to 1/2,so that the total energy of each trans-mitted symbol is to be one).Therefore no matrix in-version is involved in the calculation of ^h [i ]and the computation is also numerically more stable.With the updated channel estimation information,we can get a new estimation of X .The iteration will be repeated until a stopping criterion is reached.2.2 DFT-based initialization of CSIThe initial CSI ^h [0]is obtained by the DFT-based channel estimator with the aid of the optimum training seguence [6].If we denote P as the N X 1training se-guency domain,the training seguence for the second transmission antenna P 2is formulated as the following rule:P 2=FP 1(11)where F denotes an N X N diagonal matrix with eachdiagonal entry egual to exp -j 2!n "()Nfor the fre-guency domain index 0S n <N and "is egual to N /2.Without loss of generality,we assume that the number of subcarriers N is even and ">L .During the training period Eg.(1)can be ex-pressed asY =22i =1diag(P i )"i + (12)Then the sum of the two antennas*freguency response can be estimated by^H total (n )=Y (n )P 1(n )=H 1(n )+H 2(n )exp -j 2!n "()N +Z (n )P 1(n )0S n <N (13)Since the training seguence is designed specific-ally for each antenna,the individual channel impulse response is easily distinguished in the time domain.Then the freguency domain estimated in Eg.(13)will be transformed into the time domain by N -point ID-FT,^h total (k )=h 1(k )+h 2(k -")+~Z (k ) 0S k <N(14)If the duration of channel impulse response does not exceed L and ">L ,the initial CSI ^h [0]can be ex-pressed by^h [0]={^h total (0),.,^h total (l ),.,^h total (L -1)}T{^h total ("),.,^h total ("+l ),.,^h total ("+L -1)}[]T(15)Then the noise variance and R h can be obtained by^!2Z =2"-1k =L ^h total (k )2+2N -1k ="+L ^h total (k )()2N N -2L(16)^R h =^h [0](^h [0])H(17)2.3 Adaptive training schemeThe iterative receiver uses the current decisions to update the CSI and then it is applied in symbol de-tection for the next OFDM symbol.Although the error correcting codes and iterative algorithm can alleviate error propagation caused by erroneous decisions,the more effective methods should be considered.In Ref.[8],the suppression of error propagation is achieved by inserting the training seguence periodi-cally (with one training block sent per 10OFDM921Iterative receivers for SFBC-OFDM systems with acaptive training schemeimproved algorithm,except for the first training se-guence,the other training seguences are transmitted according to the reguest of the decoder.The maximum iteration number is imax.Once the algorithm does not converge during two successive OFDM symbols afterimaxiterations,we consider that error propagation oc-curs.Then one more training seguence is reguired to be sent by the transmitter to update the initial channel estimation.The simulation results show that this sim-ple improvement can prevent error propagation effec-tively,while keeping low system overhead.2.4 Algorithm summaryThe proposed algorithm is summariZed as fol-lows:Step1Set i=0,transmit the optimum training seguence and then perform the DFT-based channel es-timation.The initial CSI^h[0],noise variance and Rh are obtained by Egs.(15)to(17).Step2During the data transmission mode, given^h[i],obtain the estimation of^X[i+1]according to Eg.(8).After deinterleaving and error correcting,the detected data is re-encoded to reconstruct^S[i+1].Step3Feed^S[i+1]to the channel estimator to update the CSI according to Eg.(9)and set i=i+1.Step4If i<imaxand the iteration does not con-verge,i.e.^X[i]F^X[i-1],go to step2.If i S imaxand the iteration converges,i.e.^X[i]=^X[i-1],stop the itera-tion,set i=0,^h[0]=^h[i],go to step2and begin a newiteration process for the next OFDM symbol.If i=imaxand iteration does not converge,stop the iteration and check the decoding state of the previous OFDM sym-bol.If the algorithm does not converge during the pre-vious OFDM symbol after the imaxiteration,the error propagation is predicted so go to step1;otherwise set i=0,^h[0]=^h[i],go to step2and begin a new iteration process for the next OFDM symbol.3 Simulation ResultsThe proposed iterative algorithm is applied in an SFBC-OFDM system with2X1antenna configura-tion.The carrier freguency band is at2.4GHZ and the bandwidth is800kHZ.The total number of the OFDM subcarrier is N=128;the guard interval is L=16. GPSK modulation is applied.A rate1/2convolutional code is applied as the error correcting code and the space-freguency block code proposed in Ref.[5]is used in our simulation.A four-tap Rayleigh fading channel has been investigated and vehicle speed is set to be100km/h.We assume channel parameters corre-sponding to different transmission and reception anten-na pairs are independent but with the same statisticalIn each independent simulation,1000OFDMsymbols are transmitted.The performance averaged over independent simulations is evaluated.To illustrate the performance improvement of the proposed algo-rithm,two training schemes have been investigated: the scheme with the fixed training interval and the im-proved adaptive training scheme.In the first experiment,we consider the SFBC-OFDM system with the fixed training interval,the training interval Mtis set to be10,i.e.one training block is sent every10OFDM blocks.Fig.2shows the average bit-error-rate(BER)of the proposed iterative algorithm.It is shown that our proposed iterative re-ceiver with imax=2has SNR gains of about2dB per-formance improvement with respect to the case withno iterations and the increase of imaxshows some mod-erate improvement.So the maximum iteration time is set to be2in the following experiments,which makes the algorithm have low complexity and short latency.Fig.2 BER performance with training interval Mt=10 Fig.3provides the comparisons of BER perform-ance and the overhead ratio of the training seguence between the scheme with fixed training interval and the adaptive training scheme with error propagationdetection,which is denoted by DEP imax=2.The over-head ratio of the training seguence is calculated as the ratio of the training block number to the number of OFDM blocks transmitted.The simulation results demonstrate that for systems with a fixed training in-terval,the BER performance decreases strictly as the training interval increases.Most of this degradation is due to more error propagation caused by the longer training interval.It is seen that the BER performance of the receivers with the improved training scheme outperforms the systems with a fixed training interval significantly.The simulation results also show that for the systems with error propagation detection,the over-head of the training seguence drops sharply as the SNR increases and keeps lower over the concerned031Jing Ya,Xu Xiaodong,Chen Ming,and Cheng ShixinFig.3 Comparison of BER performance and overheadratio with imax=24 ConclusionIn this paper we have presented a low complexity iterative symbol detection and channel estimation al-gorithm for SFBC-OFDM systems.The methods to es-timate the noise variance and covariance matrix of the channel response have been derived.An adaptive training scheme to suppress the error propagation is proposed and applied to improve the BER perform-ance and keep low system overhead.It has been shown by simulation results that the iterative algorithm brings better performance than that without iteration.Moreover the improved training scheme can prevent error propagation effectively,con-seguently improving the system performance signifi-cantly.References[l]Sampath H,Talwar S,Tellado J,et al.A fourth-generation MIMO-OFDM broadband wireless system:design,per-formance,and field trial results[J].IEEE Commun Mag, 2002,40(9):l43l49.[2]Sun S,Wiemer I,Ho C K,et al.Training seguence assistedchannel estimation for MIMO OFDM[A].In:Proc of IEEE Wireless Communication and Networking Conference[C].New Orleans,2003,1:3843.[3]Lu B,Wang X,Li Y.Iterative receivers for space-timeblock-coded OFDM systems in dispersive fading channels [J].IEEE Trans Wireless Commun,2002,1(2):2l3225.[4]Lee K F,Williams D B.A space-freguency transmitter di-versity technigue for OFDM systems[A].In:Proc of IEEE Global Telecommunications Conference[C].San Francisco,2000,3:l473l477.[5]Alamouti S M.A simple transmit diversity technigues forwireless communications[J].IEEE Journal on Selected Areas in Commun,l998,16(l8):l45l-l458.[6]Li Y.Simplified channel estimation for OFDM systemswith multiple transmit antennas[J].IEEE Trans Wireless Commun,2002,1(l):6775.[7]Kay S M.Fundamentals of statistical signal processing:estimation theory[M].Englewood Cliffs,NJ:Prentice Hall,l993.325327.[8]Du J X,Li Y.MIMO-OFDM channel estimation based onsubspace tracking[A].In:Proc of IEEE Vehicular Tech-nology Conference[C].Jeju,Korea,2003,2:l084l088.SFBC-OFDM系统中基于自适应训练机制的迭代接收机井雅徐晓东陈明程时昕(东南大学移动通信国家重点实验室,南京2l0096)摘要:研究了在空频块码正交频分复用(SFBC-OFDM)无线通信系统中适用于多径衰落信道下的迭代接收机的设计,导出了一种迭代的联合信道估计与符号检测的算法.在提出的算法中,信道估计器交替地工作于2种模式.在训练阶段,采用基于DFT的估计器估计出信道状态信息,并且采用所提出的算法估计出噪声方差和信道响应的互相关矩阵.在数据传输模式下,迭代地获得发送数据和信道状态信息.为了抑制由于符号检测中误判引起的错误传播,提出了一种简单的错误传播判定准则,并使用了一种自适应的训练机制来抑制误差传播.仿真结果显示,与传统的迭代算法相比,所提出的算法能够提供更好的误码性能,且节约了系统开销.关键词:SFBC-OFDM;迭代接收机;信道估计;自适应训练机制l3lIterative receivers for SFBC-OFDM systems with adaptive trainingscheme。