【作者】 张俊林；

【导师】 曾孝平；

【作者基本信息】 重庆大学 ， 电路与系统， 2010， 博士

【摘要】 多用户检测技术利用各个用户的扩频序列、时延、幅度和相位信息对各用户进行联合检测,能有效抑制多址干扰,充分利用上行链路频谱资源,提高系统性能和容量,是CDMA系统的关键技术之一。著者在从事“民航客机移动通信技术”课题的研究中,广泛吸取国内外的有益成果,利用子空间方法从多用户检测算法、子空间跟踪算法、空间特征信息提取等各个方面展开研究,在工程可实现的复杂度下,寻找性能优良的检测算法以提高系统性能,提出了基于子空间的改进检测算法、基于干扰子空间的检测算法以及空时联合检测的多用户检测算法。本论文的主要内容和研究成果如下:提出了基于子空间跟踪的空时盲自适应MMSE多用户检测算法。该算法利用阵列天线技术,将信号的空间特征引入信道模型,用子空间特征向量、特征值对角矩阵和空时流形矩阵参数构造多用户检测的权矢量表达式,避免了时域一维检测中,不能有效克服同信道干扰问题。建立的空时信道参数的盲估计算法,将时域检测和空域检测的信息集中反映在多用户检测权矢量表达式中,既反映信号的时间特性(包括时延、多普勒频移、快衰落等),又反映信号空间特性(如多径信号的到达方向、天线阵列几何、角度扩展等),有效提高了系统抗干扰能力。提出一种新的线性MMSE盲自适应多用户检测算法。在基于信号子空间的线性盲自适应多用户检测算法的基础上,将接收信号投影到信号子空间,求其自相关矩阵的逆矩阵,构造一种不含子空间特征值矩阵参数的检测算法,消除了近似估计特征值矩阵引入的误差。仿真结果表明,与其它子空间检测算法相比,该算法具有更低的误码率性能。提出了基于干扰子空间的线性MMSE盲自适应多用户检测算法。将所有干扰信号的自相关矩阵进行特征值分解,得到干扰子空间特征向量和特征值对角矩阵,在最小均方误差准则下,构造基于干扰子空间的盲自适应多用户检测算法,并利用快速迭代子空间跟踪算法(FAPI)估计干扰子空间特征向量。仿真结果表明,与其它算法相比,该算法具有更高的输出信干比。针对PASTd算法估计子空间特征向量,存在特征向量不严格正交的缺点,提出了基于NewPASTd算法的线性MMSE盲自适应多用户检测算法。在PASTd算法的基础上,通过对每次迭代提取的特征向量进行单位正交化处理,保证所有特征向量都相互正交,得到一种改进的子空间跟踪算法(NewPASTd),并用于多用户检测算法中的信号子空间跟踪。仿真结果表明,与基于PASTd的多用户检测算法相比,本文提出的多用户检测算法具有较快的收敛速度。在研究了多用户检测算法的基础上,用MATLAB平台对提出的各种检测算法进行仿真实验,仿真结果与理论分析具有很强的一致性,证明了提出的多用户检测算法的可行性和有效性。更多还原

【Abstract】 Multi-user detection (MUD) applies the user’s PN sequence, delay, amplitude and phase information to make combined detection in order to control multiple access interference(MAI) , make full use of uplink frequency spectrum, promote system performance and capacity, which is one of the key technology for CDMA system. In this paper, During the project of airliner mobile communication technology study and absorbing the useful achievements, the writer applied subspace method to study on Multi-user detection algorithm, subspace tracking algorithm and spatial signature extraction, studied the suitable and feasible method to enhance the performance of MUD. Based on the study, the author proposed many linear blind adaptive MUD algorithm in order to improve the system performance. The major work and contribution are as following.Proposed the space- time blind adaptive MMSE multi-users detection algorithm on the base of subspace tracking. This algorithm makes use of array antenna technology to construct space-time signal model with spatial signature. On the least mean square error criterion, used subspace eigenvector, eigenvalue diagonal matrix and space-time manifold matrix to construct weight vector form of multi-user detection and avoid that MUD cann’t effectively restrain co-channel interference in the time domain detection. Construct blind estimation algorithm of space-time signal channel parameters that there are space-time feature information in multi-users detection weight vector in order to improve the capacity of restraining MAI.Proposed one new linear MMSE blind adaptive multi-users detection algorithm. On the base of linear blind adaptive multi-users detection algorithm based on signal subspace, projected the receiving signal to signal subspace and solve inverse of autocorrelation matrix , and deduced detection algorithm without subspace eigenvalue matrix parameters and eliminated the error introduced by estimating eigenvalue matrix. Bit-error-ratio of the algorithms proposed is much lower than that of others subspace tracking algorithms.Proposed the linear MMSE blind adaptive multi-users detection algorithm based on the interference subspace. By performing the eigendecomposition of autocorrelation matrix of interference signal, author obtained eigenvector and eigenvalue matrix, and used interference subspace eigenvector, eigenvalue diagonal matrix to construct weight vector form of multi-user detection on the least mean square error criterion, and applied FAPI to estimate the interference subspace eigenvector. Simulation results show that , the Signal to interference ratio of the algorithms proposed is much better than that of others interference subspace tracking algorithms.In PASTd algorithm, when estimating subspace character vector, the vector is not strictly Orthogonal. For this question, the author proposes the linear MMSE blind adaptive multi-user detection algorithm on the base of NewPASTd. On the base of PASTd algorithm, in the process of iteration, make unit-norm orthogonalization to the eigenvector in order to make the estimated eigenvector orthogonal, and obtain an improved subspace tracking algorithm(NewPASd). Convergent rate of the algorithms proposed is much better than that of others subspace tracking algorithms. .After studying on the multi-user detection algorithm, author made simulation of several kinds of multi-user detection algorithm based on MATLAB. And the simulation result is almost the same with the theoretical analysis, which proves the feasibility and availability of the proposed multi-user detection algrithms.更多还原