【作者】 熊尚坤；

【导师】 韦岗；

【作者基本信息】 华南理工大学 ， 通信与信息系统， 2004， 博士

【摘要】 在DS-CDMA通信系统中，多用户检测用于消除系统的多址干扰和码间串扰，是一种关键的接收技术。目前多用户检测的研究大概可以分成两个方向。一个是考虑在实际系统中与其它技术的结合运用，一般使用比较复杂的数学模型。例如把多用户检测和信道编码相结合是一个重要的方向。再如考虑邻小区干扰、多码率CDMA、长码系统等。另一个方向侧重于新算法及性能的研究，一般只考虑简单的数学模型(短码直扩模型)和输入(独立同分布的输入)，注重于算法性能如误码率、稳定性、计算量等的改进。本文将注重后者。CDMA系统按信道模型的不同分为高斯白噪声信道和衰落信道、时变和时不变信道。本文针对多径衰落慢时变信道下CDMA系统多用户检测技术的理论进行研究。主要工作如下： 1．提出了一种新的基于输出最小均方误差准则的约束优化算法-LCMMSE(Linearly Constrained MMSE)算法。该算法的代价函数和约束条件都基于最小均方误差准则，故能保证最优MMSE解落在约束条件约束的空间里。本文证明了约束空间里最优解的唯一性，同时还给出了算法实现的迭代方法，并对迭代算法的收敛性进行了详细的理论分析。 2．盲多用户检测可以通过块处理算法和自适应算法来实现，自适应算法对于简化多用户检测的算法复杂度和跟踪信道的变化有着重要的意义。本文针对LCMMSE算法提出了块处理算法和自适应算法。和现有的算法相比，在各种场合下，仿真结果均显示本文算法的性能明显优于现有算法。仿真结果还显示，本文提出的块处理算法可以在几次的迭代后收敛，自适应算法也可以在数百次迭代后收敛。 3．常模算法CMA(Constant Modulus Algorithm)是一种重要的盲多用户检测算法。在多用户场合下同样需要约束检测器的搜索空间。本文对基于常模准则的多用户检测算法进行了研究，提出了一种新的约束方法，新的约束方法基于MMSE准则。仿真结果显示本文算法性能优于现有的线性约束CMA算法。 4．直接基于最小误码率准则的多用户检测技术能取得很好的误码率性能。但是现有的最小误码率代价函数是一个极不规则的非线性函数，存在很多局部极小点。约束条件的选取是一个难题。本文提出了一组新的约束条件，证明了在该约束条件的约束空间里误码率代价函数的最小点是唯一的。本文采用阻碍梯度算法实现了约束条件下基于最小误码率准则的多用检测算法。仿真结果显示本文算法收敛较快，误码率性能接近最小误码率底限。 5．由于多用户检测器的最小误码率代价函数是一个极不规则的非线性函数，存在许多的局部极小点，本文尝试用遗传算法来解决这个全局最优问题。本文采更多还原

【Abstract】 Multiple access interference (MAI) caused by code non-orthogonality and inter-symbol interference (ISI) due to the time dispersion of wireless channels are the main difficulties in wide-band code division multiple access (CDMA) systems. Multiuser detection(MUD) that suppresses the MAI as well as the ISI remains an interesting topic in the past decades. The current research of MUD focuses on two directions. One is toward practical applications. Commonly complicated models regarding practical issues are applied; such as the turbo multiuser detection, which considers channel decoding and multiuser detection jointly. The other direction focuses on new algorithm development and performance analysis. Commonly simple system models, for instant a baseband system with short spreading code, are applied. Researches in this direction generally assume the user sequences have identical independent distribution and focus on improvement in BER, stability and computation. Our researches focus on the latter. According to different channel models, CDMA systems can be divided into additive white gaussian noise (AWGN) channel and dispersive channel, time-variant and time-invariant channel. In this paper, we mainly study multiuser detection in slow time-variant dispersive channels. The contributions are in five folds.1. Constrained optimization(CO) based methods play an important role in MUD. A CO method generally includes a cost function and a constraint to confine the searching space. In this thesis, a new algorithm based on linearly constrained(LC) MMSE criterion is proposed.. The key novelty is that the cost function and the constraint are developed based on the same criterion, which guarantees that the desired optimum can be located in the confined searching space. An iterative implementation is proposed and its convergence is also discussed.2. Blind multiuser detection can be implemented by batch algorithm and adaptive algorithm. Adaptive algorithm is more feasible to trace the channel variation. In the thesis we also propose a batch algorithm and an adaptive algorithm for the LCMMSE detector. Their performances are evaluated by computer simulations, which show the proposed LCMMSE detector outperforms some typical existing detectors.3. Constant Modulus (CM) is an important criterion in MUD. Since the cost functions of CM based Algorithms are nonconvex and there exists local minima. It is更多还原