【作者】 钱斌；

【导师】 杨万麟；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2008， 博士

【摘要】 无线信道空时参数联合估计是现代无线通信中的一个重要研究课题,也是智能天线技术在第三代移动通信系统中应用所面临的主要问题和基本需求。在复杂的移动通信环境中,信号因折射、绕射和散射等原因产生多径传播现象,由此导致信号在空间与时间上扩展,造成码间干扰和同信道干扰。这些干扰越来越成为影响现代移动通信质量的主要因素。在移动通信系统中采用空时联合处理技术可以抑制多径传播、增大系统容量、提高通信质量。而信道的空时参数联合估计正是空时联合处理中的基本技术之一,并可为空时联合处理中的其他技术提供所需求的基本参数,具有重要的理论意义和实用价值。现代通信环境中,由建筑物、道路或其他因素导致的局部密集散射使得接收天线与发射终端之间常常不存在直达路线,信号传播更多地依赖于大量的非直达路线,因而导致信号在空间、时间上都产生一定程度的分布扩展,形成空时分布式多径信号(space-time distributed signals),此种情况下,基于传统点源直达波模型的空时参数估计技术不再适用,甚至可能带来估计性能的恶化。本论文针对上述问题,以通信参数为主要研究对象,在深入分析现有模型的基础上,对空时分布式多径信号(分布式信号)的波达方向(DOA)-时延联合参数估计问题作了系统性的研究。主要创新之处如下:1.对现有的空时分布式信号的接收模型和信道模型进行了细致的理论推导,并提出了一种带有观测噪声的信道模型,详细地分析了噪声对参数估计中某些技术方法(如信道的傅立叶变换和去卷积计算)的不利影响。2.基于噪声模型的分析,提出了一种新的去卷积和获得空时联合信号子空间的方法,该方法可以避免由于对信道估计数据的傅立叶变换和去卷积计算而导致的噪声方差变大,从而空时联合信号子空间受到噪声污染的问题。3.利用分布函数信息,改进了传统空时联合参数估计中的时-空-时MUSIC方法(TST-MUSIC),改写了空间滤波矩阵和时延滤波矩阵,解决了点反射源方法估计分布式信号时的失效问题。4.在搜索谱峰的参数估计方法中,提出抽取子空间的概念,通过在联合信号子空间内分别抽取不同的行来构造空间搜索矩阵和时延搜索矩阵,并利用分布的对称性信息和搜索矩阵瑞利商的性质,将复杂的多维搜索降为两个一维搜索和一个配对过程,加快了算法的计算速度,并给出另外两种快速搜索方法。5.具体分析了扩展分布对旋转不变性质的影响。给出了保证空、时流形向量相邻元素间相位上保持旋转不变性的条件,并以此为依据确定了阵元数和去卷积参数的取值范围。6.给出了两种基于旋转不变性质具有闭式解的空时参数联合估计方法:抽取ESPRIT方法和时空矩阵方法。改变了对分布式信号空时参数估计中没有闭式表达式方法的现状。7.在抽取ESPRIT算法的基础上改进了参数估计过程,提出了一种能够在一定程度上抑制非对称分布误差的参数估计方法,使算法降低了对非对称分布误差的敏感性。8.在多条路径具有相同参数时(多路同参),分析了空时参数联合估计方法中的参数兼并现象,证明了抽取ESPRIT方法对于非同参多径空、时参数估计的可能性,确定了独立多径的空时参数匹配关系。对于多路同参路径,提出了一种ESPRIT-MUSIC联合估计算法,解决了基于空间平滑技术的传统方法无法应用于分布式信号的问题,达到解参数兼并的目的。与现有的基于最大似然准则的分布式信号DOA-时延联合估计方法相比,本文提出的基于谱估计和子空间分解的方法能够有效避免多维参数估计中的复杂优化问题,降低计算复杂性,并且可以给出闭式表达式,具有不敏感于信号分布特性和分辨率高的优点。更多还原

【Abstract】 Joint space-time parameter estimation is an important research problem in modern wireless communication systems. It is also a main concern when smart antenna technology is used in 3G systems. In complicated communication environments the multipath broadcasting caused by reflection, diffraction or refraction always lead to intersymbol interference which is the main reason for communication quality degradation. As a result the use of joint space-time parameters estimation in mobile system can restrain multipath phenomena, increase system capability and improve communication quality. At the same time the space-time parameters is very important for other space-time processing technology. So joint space-time parameter estimation method has its theory significance and application merit.Because of dense reflection caused by buildings, roads or other reasons there is no direct path between transmitter and receiver. The Not-Line-of-Sight(NLOS) propagation may cause angle spread and delay spread in transmitting signals. So traditional point-based model is not useful anymore and may bring communication quality deterioration. So it is very important to theorize and analyze distributed signal model in both space and time and is the main concern of this dissertation. Based on existing models systematic research is done to estimate DOA(direction-of-arrival) and delay parameters jointly. Several valuable and important achievements of this thesis which bring new ideas are indicated as follows:1.The deduction of space-time distributed channel model is given by considering observation noise. The performance character versus noise is anatomized carefully. The adverse influence of noises is analyzed for some channel transform technology (eg: fourier transform and deconvolution).2.A new deconvolution and joint signal subspace acquisition method is proposed and has the advantage of noise suppression.3.The traditional TST-MUSIC(multiple signal classification) method is modified using space and time distribution function. So it can be used when channel signal is distributed.4.The concept of extract subspace is put forwarded. The space search matrix and delay search matrix are constructed by extracting rows from joint space-time signal subspace. The 2-dimension search process for distributed channel estimation is decomposed into two 1-dimension search and a pairing process by using the symmetric property of space and delay distribution.5.The rotation invariant property is analyzed by considering distributed parameters in space and delay domain. Then the condition for rotation invariant property in phase is given.6.Two close-form method are given to DOA-delay joint estimation of distributed channel signals: the extract-ESPRIT and space-time ESPRIT.7.A post treatment process based on extract-ESPRIT is proposed by which estimation error can be retrained when actual signal mismatch the supposed symmetric property.8.When several paths have same parameters in space or delay the parameter merger problem is considered thoroughly. And the possibility of estimating independent path parameters by extract-ESPRIT method is proven. For those paths which have same parameters with others, a ESPRIT-MUSIC method is given while traditional space-smoothing-based decoherence technology does not work in distribution environments.In contrast with the existing maximum likelihood method, methods in this dissertation can give more accurate estimations and use less computational complexity. Those spectra methods and subspace decomposition methods proposed are insensitive to signal distributions as well.更多还原