Research On Blind Channel Estimation Of OFDM Systems

Orthogonal frequency-division multiplexing(OFDM)technology is widely used in wireless communication because of its advantages of high data transmission rate and high spectrum utilization.By the combination of frequency diversity and time diversity,OFDM technology can not only greatly improve the channel capacity and transmission rate of the system,but also resist multipath fading and interference effectively.In practical applications,OFDM systems usually need channel state information for coherent demodulation,so channel estimation is an important task for the receiver design of MIMO-OFDM systems.The existing channel estimation methods for OFDM systems can be summarized as blind(semi-blind)channel estimation and non-blind channel estimation.Non-blind channel estimation methods used to send the structure of the training sequences or insert pilots,whereas blind channel estimation methods do not need to send the structure of the training sequences or insert pilots,channel information can be estimated online only by received signals,which can improve the spectral efficiency.We studied on the blind channel estimation problem of MIMO-OFDM systems in this dissertation.The main contributions are as follows:1.To solve the channel estimation of SISO-OFDM systems,this thesis proposes a precoding based modified joint blind channel estimation algorithm.By introducing a precoding matrix on the transmitted symbols,the correlation between the receiving symbols can be improved,so that the channel can be estimated by using this correlation at receiver.The structural characteristics of the autocorrelation matrix R of received signal has been analyzed in detail.Firstly,the diagonal elements of R are used to obtain the initial channel estimation,and then the initial channel parameters are recursively by the lower triangular elements of R to obtain the final channel parameter estimation.The computational complexity of this algorithm is analyzed in this thesis,its complexity is obviously reduced compared with the subspace blind channel estimation algorithm.Finally,the simulation results show that the proposed modified joint blind channel estimation algorithm can achieve better accuracy than the existing blind channel estimation algorithm in high SNR case.2.To solve the channel estimation of MIMO-OFDM systems,a blind channel estimation algorithm based on non-redundant block precoding is proposed.The block coding matrix are introduced for two transmitted symbols,then singular value decomposition for two diagonal sub-blocks of autocorrelation matrix of received signals can be done to obtain two kinds of MIMO channel estimation,respectively.By mixing the two estimations together,the final estimation of channel impulse response can be obtained.Finally,the selection principle of fusion coefficient is given,which can guarantee the optimal channel estimation and minimizing the bit error rate at the same time.Simulation results show that the proposed algorithm can obtain better performance than other existing blind channel estimation algorithms.3.To solve the channel estimation of SIMO-OFDM systems,a blind channel and symbol joint estimation method based on parallel factor analysis is proposed.A parallel factor analysis model of receiving signals is established,and then the right singular vectors can be obtained by singular value decomposition of the received signal matrix.Combining with the discrete Fourier transform matrix,a nonsingular matrix can be constructed.Thus the channel estimation and signal detection can be obtained by the nonsingular matrix.The proposed algorithm is a closed solution method,which has lower computational complexity than the traditional method.Simulation results show that the proposed algorithm has better channel estimation performance than the traditional methods.4.To solve the adaptive blind channel tracking problem for time-varying MIMO-OFDM systems,two algorithms are proposed:(1)Recursive least squares blind channel tracking algorithm,an index function of receiving error is designed in the form of exponential window least squares,by minimizing the index function,the channel estimation can be obtained;(2)Synchronization diagonalization blind channel tracking algorithm,a nonsingular matrix is constructed by singular value decomposition of the windowed observation matrix,and then the channel estimation can be tracked by updating the nonsingular matrix.Simulation results show that the two proposed blind channel tracking algorithms have faster convergence speed and smaller channel tracking error than traditional blind channel tracking algorithms.