MIMO-OFDM系统压缩感知稀疏重构算法研究

发布时间：2018-10-25 07:17

【摘要】：模拟通信系统到数字通信系统的发展是移动通信领域的重大转变,随着人们对高速率、大容量数据传输的需求,通信传输技术也在不断发展。正交频分复用(OFDM)技术有效地提高了信道利用率,多输入多输出(MIMO)技术充分利用空间资源,提高信道容量。但是,在实际信道中,OFDM技术对无线通信系统的时偏和频偏都异常地敏感,尤其MIMO技术的应用将增加这种问题的严重性。若系统接收端接收到精准的无线信道特征可以减少失真,所以信道估计技术尤为重要。针对现实传输信号系统存在的一些问题,本课题先对传统的基于导频估计算法进行研究,再对基于压缩感知(CS)的估计算法进行研究。主要的研究工作和创新性的研究成果如下:(1)基于离散傅里叶变换(DFT)的OFDM信道估计研究。针对DFT仅改善CP外的样本点噪声,没有考虑到CP内噪声的问题,采用一种新的改进阈值的估计算法。该算法的思想是首先对CP内样本点的能量取其平均值,然后对CP外样本点能量升序后取合适区间上的能量的均值,再取两者之和作为本文的阈值。仿真结果证明了本文采用的新阈值方法有效地提高了系统估计性能。(2)对于CS重构算法的MIMO-OFDM系统信道估计研究。针对传统的信道估计算法若要达到精确度较高的信道特征,需要大量导频,导致频谱利用率低、导频污染严重的问题,利用MIMO-OFDM系统的时域上的稀疏特性,采用CS重构信号方法对MIMO-OFDM系统进行性能估计,仿真结果表明在获得相同的精确性性能时,采用CS重构信号方法需要的导频数远远少于采用传统导频参数的估计方法。(3)对于梯度追踪(GP)的MIMO-OFDM稀疏信道估计方法研究。针对正交匹配追踪及其改进算法在处理大规模数据时,由于需要处理大量最小二乘计算,导致计算复杂度较高的问题,采用梯度追踪重构方法对MIMO-OFDM系统进行信道性能估计。该算法采用梯度思想选择原子,避免计算最小二乘,在保证信道的估计准确性同时能减少计算复杂度,从而提高系统估计性能。仿真结果表明基于GP方法的估计运算复杂度比OMP方法有所降低。
[Abstract]:The development of analog communication system to digital communication system is a great change in the field of mobile communication. With the demand of high speed and large capacity data transmission, communication transmission technology is also developing. Orthogonal frequency division multiplexing (OFDM) technology can effectively improve channel utilization. Multi-input and multi-output (MIMO) technology makes full use of space resources and improves channel capacity. However, in the actual channel, the OFDM technology is very sensitive to both the time offset and the frequency offset of the wireless communication system, especially the application of MIMO technology will increase the severity of this problem. If the receiver receives accurate wireless channel features, the distortion can be reduced, so channel estimation is particularly important. Aiming at some problems existing in the real transmission signal system, this paper first studies the traditional pilot estimation algorithm, and then studies the compressed perceptual (CS) estimation algorithm. The main research work and innovative research results are as follows: (1) OFDM channel estimation based on discrete Fourier transform (DFT). Aiming at the problem that DFT only improves the noise of sample points outside CP and does not take into account the noise in CP, a new improved threshold estimation algorithm is proposed. The idea of this algorithm is to take the mean of the energy of the sample points in CP first, then to take the mean value of the energy of the appropriate interval after the ascending order of the energy of the sample points outside the CP, and then to take the sum of the two as the threshold of this paper. Simulation results show that the new threshold method proposed in this paper can effectively improve the performance of system estimation. (2) the channel estimation of MIMO-OFDM system based on CS reconstruction algorithm is studied. In order to achieve the channel characteristics with high accuracy, the traditional channel estimation algorithm needs a large number of pilots, which leads to low spectral efficiency and serious pilot pollution. Therefore, the sparse characteristic of MIMO-OFDM system in time domain is utilized. The CS reconstruction signal method is used to estimate the performance of MIMO-OFDM system. The simulation results show that the same accuracy performance is obtained. The number of pilots needed to reconstruct the signal by CS is much less than that by using the traditional pilot parameter estimation method. (3) the MIMO-OFDM sparse channel estimation method with gradient tracking (GP) is studied. In order to solve the problem of high computational complexity caused by the need to deal with a large number of least square calculations when dealing with large scale data, gradient tracking reconstruction method is used to estimate the channel performance of MIMO-OFDM systems. The algorithm adopts gradient method to select atoms and avoids the calculation of least squares. It can reduce the computational complexity while ensuring the accuracy of channel estimation and thus improves the performance of system estimation. Simulation results show that the estimated computational complexity based on GP method is lower than that of OMP method.
【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.3;TN929.53

【参考文献】