基于压缩感知的大规模MIMO导频设计与信道估计方法研究

发布时间：2018-03-25 14:49

本文选题：大规模MIMO　切入点：信道估计　出处：《东南大学》2017年硕士论文

【摘要】：随着多媒体业务、物联网以及大数据的发展,未来移动通信系统的数据传输速率需要提高上千倍甚至上万倍,现有的通信技术已经无法满足这种需求。大规模多输入多输出(MIMO,multiple-input and multiple-output)系统由于能够深度利用空间自由度,显著提高通信系统数据传输速率,被认为是下一代移动通信的核心技术。然而,大规模MIMO系统的性能很大程度上取决于基站侧获取的信道状态信息(CSI,channel state information)的质量,传统的基于正交导频的信道估计方法需要的导频开销和反馈开销都将随着基站侧天线数目的增多而增大,这无疑将限制大规模MIMO带来的性能提升。因此,本文针对信道估计快速算法,导频设计等问题展开深入地研究。首先,本论文研究了 MIMO系统中传统下行信道估计方法,包括最小二乘(LS,least square)、缩放最小二乘(SLS,scaled least square)以及最小均方误差(MMSE,minimized mean square error)等方法,详细地阐述了这些方法的工作原理与过程,分析了相应方法下的最优导频结构,给出了各种方法的均方误差性能,说明了基站侧天线数目对每种方法的误差性能以及导频开销的影响,并详细介绍了这些方法的改进与简化。其次,针对大规模MIMO系统下传统信道估计方法导频开销过大、计算复杂度过高的问题,提出了基于压缩感知的大规模MIMO波束域信道估计方法。大规模MIMO信道在波束域具有稀疏性,信道能量主要集中在少数几个波束上,因此可利用该特性减少导频开销。根据波束域稀疏特性,本文提出了基于压缩感知的波束域信道估计方法,给出了满足约束性等距条件(RIP,restricted isometry property)的非正交导频设计,并提出了修正子空间追踪(M-SP,modified subspace pursuit)算法来解决近似稀疏信号恢复的问题。仿真结果表明该方法的性能显著优于传统LS信道估计的性能,并且可以有效减少导频开销。最后,本论文重点研究了大规模MIMO正交频分复用(OFDM,orthogonal frequency division multiplexing)系统下的信道估计方法和导频设计,提出了基于压缩感知的大规模MIMO-OFDM波束-时延域信道估计方法。大规模MIMO-OFDM系统在波束域具有稀疏性,同样在时延域,信道抽头也是稀疏的,其中小部分的抽头系数占据了大部分的信道能量。因此,综合利用信道在波束-时延域的稀疏特性,提出了波束-时延域信道估计方法,在波束-频率域设计了对应的导频-移相Zadoff-Chu导频(PSZCPs,phase shift Zadoff-Chu pilots),并推导了对应波束-时延域的测量矩阵,同时提出了利用统计信道状态信息的MMSE-SP算法。仿真结果表明本文提出的波束-时延域信道估计方法可以有效减少导频开销并显著提升基于移相正交导频(PSOPs,phase shift orthogonal pilots)的信道估计的性能,该方法只需要一个OFDM符号,而传统的基于PSOPs的信道估计需要多个OFDM符号。另外,MMSE-SP算法相比于SP算法取得了较大的性能增益,并且减少了对精确稀疏度的依赖。
[Abstract]:With the development of multimedia services, networking and big data, the data transmission rate of future mobile communication systems need to improve thousands or even millions of times, the existing communication technology has been unable to meet this demand. Large scale multi input multi output (MIMO, multiple-input and multiple-output) system due to the depth of the use of degrees of freedom, improve data transmission the rate of communication system, is considered to be the core technology of next generation mobile communications. However, the performance of large-scale MIMO system depends largely on the base station side gets the channel state information (CSI channel, state information) the quality of the traditional orthogonal pilot channel estimation method requires the pilot overhead and feedback overhead with the base station antenna number increasing. Based on this, undoubtedly will limit the performance of large scale MIMO brings improvement. Therefore, this paper aims at the channel Fast estimation algorithm, in-depth study of pilot design issues. Firstly, this paper studies the traditional downlink channel estimation method in MIMO system, including the least squares (LS, least, square) (SLS, scaled least zoom square square) and minimum mean square error (MMSE, minimized mean square error) and other methods, detailed describes the working principle and process of these methods, analyzed the corresponding optimal method under the pilot structure, the mean square error performance is given a variety of methods, shows the number of base station antenna for each method of the error performance and the effect of pilot overhead, and introduces the improvement and simplification of these methods. Secondly, the traditional channel estimation method for large-scale MIMO system pilot overhead is too large, the computational complexity problem, and proposed a method to estimate large-scale MIMO beamspace channel based on compressed sensing mass. The MIMO channel is sparse in beam domain, channel energy mainly concentrated in a few beams, so you can use this property to reduce the pilot overhead. According to the beam domain sparsity estimation method is proposed in this paper, the beam domain channel based on compressed sensing, given satisfy the restricted isometry (RIP, restricted isometry property) - orthogonal pilot design, and puts forward the modified subspace pursuit (M-SP modified, subspace pursuit) algorithm to solve the approximate sparse signal recovery problem. Simulation results show that the performance of this method is significantly better than the traditional LS channel estimation, and can effectively reduce the pilot overhead. Finally, this paper focuses on the study of large scale MIMO orthogonal frequency DivisionMultiplexing (OFDM orthogonal, frequency division multiplexing) method and design of pilot channel estimation system, proposed a large-scale MIMO-OF based on compressed sensing DM beam delay domain channel estimation method. The large scale MIMO-OFDM system is sparse in the beam domain, at the same time delay domain, channel taps is sparse, the tap coefficients of small part of the channel occupied most of the energy. Therefore, the comprehensive utilization of the channel delay characteristics of the beam in sparse domain, a method is proposed to estimate the beam delay in the beam domain channel, frequency domain pilot design corresponding to the pilot phase Zadoff-Chu (PSZCPs, phase shift Zadoff-Chu pilots), and the measurement matrix corresponding to beam delay domain is derived, and put forward the MMSE-SP algorithm using statistical channel state information. The simulation results show that the proposed beam delay domain channel estimation the method can effectively reduce the pilot overhead and enhance phase shift orthogonal pilot based on (PSOPs, phase shift orthogonal pilots) the performance of channel estimation, the method needs only one OFDM Symbol, and the traditional PSOPs based channel estimation needs multiple OFDM symbols. In addition, MMSE-SP algorithm achieves a large performance gain compared with SP algorithm, and reduces the dependence on precise sparsity.

【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.3

【相似文献】