【作者】 王燕；

【导师】 张立毅；

【作者基本信息】 天津大学 ， 信号与信息处理， 2018， 博士

【摘要】 随着人们对无线通信需求的快速增长,现代移动通信系统标准中对快速移动性、高速传输速率的要求也不断提高,在众多的4G通信标准中都需要系统可以提供移动速度在350km/h以上的宽带通信服务。正交频分复用(OFDM)技术和多输入多输出(MIMO)技术都是4G通信系统中的核心技术。OFDM采用多载波调制技术,在提高频带利用率的同时,也可以很好的抑制信道时延和衰落对系统的影响。MIMO采用空间分集技术,提高了系统的容量。由于OFDM对同步误差较为敏感,而信号在OFDM的传输过程中,会受到多径衰落的影响,会产生不同程度的符号间干扰(ISI);同时由于接收端的快速移动性,会引起多普勒频移,从而产生子载波间干扰(ICI);再加上系统采用了MIMO技术,使MIMOOFDM系统的无线信道环境变得更为复杂。因此,如何获得实时、精确的信道估计信息就成为MIMO-OFDM系统中研究的热点之一。传统的信道估计算法主要分为基于导频的信道估计和盲信道估计两类,而它们均是将信道作为密集多径信道进行估计的,并没有考虑到宽带信道的稀疏性及稀疏结构的问题,因此不适用在宽带信道的估计中。针对上述问题,本论文将新兴的信号处理理论——压缩感知(CS)应用到宽带信道估计中,在分析了宽带无线信道具有稀疏性及特殊的稀疏结构的基础上,主要研究了基于压缩感知的OFDM系统和MIMO-OFDM系统的频率选择性信道、时延-频移双选择性信道估计算法并提出了基于修正的压缩信道跟踪算法。论文的主要工作内容和创新点包括以下四个方面:1.提出了用于频率选择性信道的基于分簇特性正则化自适应匹配追踪压缩信道感知算法(CRAMP)。论文首先分析了OFDM系统的频率选择性衰落信道特征,得出了具有分簇结构的稀疏信道模型。将压缩感知技术应用到信道估计中,即使用压缩感知算法恢复信道状态信息——压缩信道感知。在此基础上根据信道的分簇特性,提出了一种基于分簇特性的正则化自适应匹配追踪压缩信道感知算法。该算法将块-压缩感知理论应用到信道估计中,并采用正则化方法对恢复的信道参数进行二次筛选。通过仿真实验表明,该算法可以在使用较少的导频下获得较高的信道重构精度,与传统的压缩信道感知算法相比具有更高的重构精度。2.提出了利用2D-DFT作为观测矩阵的应用于时延-频率双选择性信道的压缩信道感知算法。论文在分析了OFDM系统的频率选择性衰落信道特征的基础上,将等效稀疏信道模型扩展到时延-频率双选择信道,并根据建立的模型,将压缩信道感知应用到时延-频率双选择信道状态信息重构中。提出了一种将2D-DFT矩阵作为观测矩阵的压缩感知算法。该算法根据信道状态信息的时频相关性,将观测矩阵分成时频网格块,再采用正则化匹配追踪算法来恢复信道状态信息。仿真实验表明,该算法在应用相同导频的情况下,可以比单一方向的块-压缩信道感知算法获得更好的重构精度。3.提出了基于分布式压缩感知的压缩信道感知算法。在分析MIMO-OFDM系统信道特征的基础上,建立了基于MIMO-OFDM系统的双选择稀疏信道模型。提出了基于分布式压缩感知的压缩信道感知算法,该算法在应用信道分簇特性的同时,在算法每次迭代时加入阈值权重参数,能自适应调整选取原子的数目,并可以根据残差自动调整阈值权重参数。该算法对快衰落信道下的MIMO-OFDM系统能以很高的概率对信道进行重构,且比多天线系统中对信道状态信息进行逐一重构的传统压缩信道感知算法的运算时间要低很多。4.提出了基于MOD-CS的压缩信道跟踪算法。由于先前研究的压缩信道感知算法中并没有对以前得到的信道估计状态信息加以利用,因此,本文提出了一种信道跟踪算法。该算法利用信道的近似连续稀疏性,使用修正压缩感知(Modified-CS)的概念,跟踪信道在一段连续的几个符号块的信道状态估计值,将其作为下一次信道估计的先验信息提供给信道估计器,即达到了信道跟踪的目的。这种方法可以产生一个附加的性能增益,可以显着降低了该方法的计算复杂度。更多还原

【Abstract】 With the growing demand for wireless communications,the requirements for fastmoving,high-speed transmission rate is improving in modern mobile communication system standard.And further the 4G communication system standards are required to provide broadband communication services of the 350km/h or more moving speed.Both OFDM and MIMO are core technologies of 4G communication systems.How to get real-time,accurate channel estimation information has become one of the hot research topics for MIMO-OFDM systems.The traditional channel estimation algorithms to the dense multipath channel estimation,and don’t take into account the sparsity and sparse structure of the wideband channel not to be suit for the estimation of the wideband channel.According to above problems,this thesis applies compressed sensing theory in wideband channel estimation.Based on the analysis of sparsity and special sparse structure of wideband wireless channel,it mainly researches the frequency selective channel,and time delay-frequency shift dual selective channel estimation algorithm based on compressed sensing OFDM and MIMO-OFDM systems,and presents the channel tracking algorithm based on modified compressed sensing.The main content and innovation of the thesis include the following four aspects:Firstly,a compressed channel sensing algorithm of adaptive regular matching pursuit(CRAMP)based on clustering feature is presented.The thesis analyzes the frequency-selective fading channel characteristics for OFDM system,and achieves sparse channel model with cluster structure.According to the cluster feature of channel,it presents a compressed channel sensing algorithm.The algorithm applies blockcompressed sensing theory to the channel estimation,using regularization method to recover the channel parameters.Simulation results show that the algorithm can obtain the higher channel reconstruction accuracy with the fewer pilots.Secondly,a compressed sensing algorithm for the delay-frequency dual selective channel is proposed.According to the established model,the compressed channel sensing is applied to the reconstruction of time delay-frequency dual selective channel.This thesis proposes a compressed sensing algorithm recording 2D-DFT matrix as the observed matrix.The algorithm,according to the frequency correlation of channel state information,divides the observed matrix into grid blocks of time-frequency,then recover the channel state information with the regularization matching pursuit algorithm.Simulation results show that in the case of applying the same pilot,the algorithm can achieve better reconstruction accuracy than the channel sensing algorithm of block compression of one direction.Thirdly,a compressed sensing algorithm based on distributed compressed sensing is proposed.Based on the analysis of the channel characteristics of MIMO-OFDM,this thesis proposes a compressed sensing algorithm based on distributed compressed sensing,which can add the parameter on threshold weight in each iteration with the application on channel clustering characteristics,adaptively adjust the number of selected atoms and automatically adjust the parameter on threshold weight according to residuals.It can also realize channel reconstruction by high probability for fast fading MIMO-OFDM and cost less operation time than traditional algorithms,when it individually reconstructs channel state information for a multiple antenna system.Finally,the channel tracking algorithm based on modified compressed sensing is proposed.This thesis presents an algorithm,using approximately continuous sparsity in channel and Modified-CS concept,tracking the channel state estimation in a continuous several symbols block which is provided to the channel estimator as a priori information at the next channel estimation to reach the aim of the channel tracking.This method can produce an additional performance gain.更多还原