【作者】 何修富；

【导师】 廖桂生；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2009， 博士

【摘要】 正交频分复用（OFDM）系统,广泛应用于无线通信系统中，由于它具有频带利用率高、抗多径干扰能力强等特点，用于数字音频广播（DAB）、数字视频广播（DVB）、非对称数字用户线（ADSL）,以及无线本地环路(WLAN)和第四代移动通信。在OFDM通信系统中，无线信道具有时变性，存在多径干扰和频率选择性衰落，若系统采用差分调制，接收端不需要跟踪估计时变信道，但存在着3db的信噪比损失；采用相干调制，系统可以采用任意星座点信号调制，接收端要准确地恢复发送信号，就要求估计出信道传输特性。本文研究了OFDM通信系统的信道估计方法，主要的工作如下：(1)对于循环前缀的正交频分复用（CP-OFDM）通信系统,针对信道的传输矩阵可以分解为延时矩阵和幅度矢量的乘积，提出了采用投影子空间跟踪（PAST）算法估计延时矩阵，基于LMS自适应算法估计幅度矢量，从而实现了利用导频信号对信道矩阵的自适应估计。同时，本文还提出了将Kalma n算法用于信道估计，用导频信号获得的信道估计作为迭代算法的初始值，运用已解调的数据和新接收的信号自适应更新Kalma n算法系数，实现了对实际信道特性的自适应跟踪估计，提高了跟踪能力。文中还提出了采用余弦变换处理导频信号的信道最小二乘估计结果，进一步提高信道估计的精度，抑制干扰的影响。(2)对于置零OFDM(ZP-OFDM)系统，对接收数据矩阵采用子空间分解获得噪声子空间矢量，由噪声子空间矢量构成Hankel矩阵，各个矢量对应的Hankel矩阵积，然后求和形成新矩阵W,W矩阵的最小奇异值对应的矢量与信道冲击响应成比例。提出了基于RLS和LMS算法对接收数据矩阵实现在线自适应噪声子空间分解，采用迭代算法实现信道特性的自适应盲估计；解决了信道盲估计的模糊系数的确定问题，通过计算机仿真，验证了算法实现的可行性。(3)本文还提出了将独立分量分析(ICA)算法运用到CP-OFDM系统的信号盲分离。直接对接收信号矩阵作白化处理，运用CP-OFDM信道矩阵的卷积结构特征，对接收信号矩阵作循环移位处理，通过ICA迭代算法，求非线性代价函数的最小值来实现信号的盲分离。算法中还采用归一化处理，克服了ICA盲算法中存在的模糊系数问题。(4)在静态信道情况下，对OFDM通信系统的误码率作了数学分析；还分析了OFDM通信系统，数据采用DPSK调制，存在多普勒频移的衰落信道下，系统的误码率表达式；同时，本文提出了运用数学方法估算MIMO-OFDM系统采用格形码调制下的系统误码率。对系统在慢衰落和瑞利衰落信道情况下的误码率作了计算机仿真。更多还原

【Abstract】 Orithogona l frequency division multiplexing (OFDM) is becoming widely appliedin wireless communication system due to its high rate transmission capability withhigh bandwidth efficiency and its robustness with regard to multi-path fading.It hasbeen used in digital audio broadcasting (DAB) system.d igita l video broadcasting(DVB) system,d igital subscriber line (DSL) standards,a nd wireless LAN standardssuch as European equiva lent HIPRLAN/2.It has also been proposed as the coretechnique for the fourth-generation (4G) wireless mobile communications.The use of differentia l phase-shift keying (DPSK) in OFDM system avoids needto track a time varying channel;it limits the number of bits per symbol and results in a3 db loss in signa l-to-noise ratio (SNR),coherent modulation allows arbitrary signa lconstellations,but efficient channel estimation strategies were required for coherentdetection and decoding. The channel estimation meathods for OFDM system aredicussed. The ma in contents are summa rized as follows:1）The first part of the thesis focuses on the algorithms for channel estimationbased on comb-type pilot in the cyclic prefixing (CP) OFDM system,the channelmatrix can be resolved for postpone matrix and range vector of product,the projectionapproxima tion subspace tracking (PAST) algorithms is introduced to estimationpostpone matrix,the least minimum square (LMS) algorathm is used to estimate rangevector,the channel matrix that corresponde to the pilot signa l frequency channelrespondence is carried out,then,interpolation is used to estimate other carrier wavefrequency channel respondence.Moreover,this text returns kalma n algorithm used forthe channel estimation and used to lead pilot signa l acquires of the channel estimationis the initia l value of the kalma n iterative algorithm ,usege decoded data and latelyreceivedsigna l from orientation renewa l kalma n coefficient,rea lization to actualchannel respondence.2）For place zero OFDM (ZP-OFDM) system,to receive the data matrix adoptionsubspace the decomposition acquire a noise subspace vector,from noise subspace vector constitute Hankel matrix , the sum of the each vector correspond ing Hankelmatrix inner product is formated the new matrix W,the minimum and strange value ofW matrix corresponde to vector is direct ratio channel impact respondence.Studiedapplied recursive least squares (RLS) and LMS in the mea ntime the algorithmresolves towards receiving data matrix to carry out on-line from orientation noisesubspace and adopt iterative algorithm carry out the channel respondence of thechannel blind estimatio n;still discussed the misty coefficient of channel blindestimation is really detemined,imitates through computer and verifies the possibility ofalgorithm realization.3）The independent component analysis(ICA) algorithm is made use to the blindsigna l separation of CP-OFDM system,the receiving signa l matrix is directly madewhite processing,pass ICA iterative agorithm,seek ing the mnimum value of thenonlinearity price function carried out the blind separation of signa l.still discussed inthe mea ntime content of the usage CP-OFDM channel matrix a convolution structurecharacteristic and circulates to move a processing towards receiving signa lmatrix,usa ge ICA algorithm carry out the blind separation of signa l.A norma lizationprocessing is adopted to overcome the misty coefficient problem in the blind algorithm.4）At the flat fading channel under the circumsta nce,the bit error rate(BER) ofOFDM system was made mathematics analysis;still discussed OFDM correspondencesystem,adopt DPSK to modulate over the frequency-selective channel due to theDoppler spreading,deduced the BER of system expression;In the mea ntime, the BERof MIMO-OFDM system adopted the trellis-coded modulation(TCM) is studied.TheBER of system over the flat-fading and rayleigh-fading channels is simulated.更多还原