Research On OFDM Channel Estimation Based On Compressed Sensing

In the conventional signal processing theory, the Nyquist sampling frequency of the signal requires not less than twice the highest signal bandwidth, in order to restore the original signal without distortion, but to sampling, storage, transmission and processing of band signals to a lot of pressure, especially in the field of OFDM systems and UWB system. Compressed sensing theory suggests that, when certain conditions are met, can be less than twice the highest signal frequency bandwidth of the signal at the same sampling and compression, then the receiving end use appropriate reconstruction algorithm accurately recover the original signal, not only breaking the Nyquist limit, but also saves a lot of signal processing resources.This article studies channel estimation of OFDM system in the field of compressed sensing theory to analyze the correlation algorithm performance and improvements.Firstly, the article describes three basic questions of compressed sensing theory: one is the sparse representation of the signal, and the second is the design of the measurement matrix, the third is reconstruction algorithm design, which is key to reconstruct the signal. This artitle focuses on reconstruction orthogonal matching pursuit(OMP), compare the performance of the OMP algorithm, least squares(LS) and the Matching Pursuit structured LS algorithms. This artitle also study the number of pilot and the shape of pilot which affect the performance of OMP algorithm.To verify when the number of pilots is 4 to 6 times the channel sparsity, the reconstruction algorithm can get better performance.Secondly, in order to further enhance the algorithm performance, this article focuses on the impact of pilot pattern on algorithm performance. According to the Restricted Isometry Property of the observation matrix, with relationship to each other, a quantitative measure of the uncorrelated observation matrix size, and design search algorithm to find the optimal location of the pilot in each pilot number, the final analysis best pilot position reconstruction algorithm performance by simulation experiments optimal effectiveness of pilot frequency position.Finally, we also studied at the channel sparsity K is unknown and how to improve the performance of stagewise orthogonal matching pursuit algorithm(StOMP). Traditional StOMP algorithm has characteristics of sparsity unknown, convergence speed, but its measurement matrix confined Gaussian random matrix, which has limited range of applications. By improving its threshold parameter, which was applied to estimate the channel areas(measurement matrix is part of the Fourier matrix), combined with the best location of the pilot study is proposed based on the improved position of the optimum pilot StOMP algorithm by experiments showed that when the threshold parameter in the range [0-1], the better the performance of the reconstruction, the range is suitable for fast varying channel or a slowly varying channel.