码分多址(CDMA)技术是第三代移动通信系统的核心技术之一。但是在CDMA系统中，由于多址干扰的存在，使系统性能受到严重影响。多用户检测(MUD)技术不但可以抗多址干扰，还可以抗远近效应和多径干扰，因此成为CDMA系统的关键技术之一。而盲多用户检测器由于具备不需要训练序列和干扰用户的先验知识等一系列优点，已经成为现在国内外的热门研究课题。本文在现有多用户检测技术研究成果的基础上，重点研究了多用户检测技术中的盲多用户检测算法。 通过研究盲多用户检测的直接序列扩频码分多址(DS-CDMA)系统模型，分析和仿真了同步DS-CDMA系统、高斯白噪声(AWGN)信道下的基于最小输出能量(MOE)准则的最小均方(LMS)和递归最小二乘(RLS)两种盲多用户检测算法。在LMS算法的基础上，通过改变步长选择方式，提出一种改进的自适应变步长LMS盲多用户检测算法，并进行了仿真，仿真结果表明该算法性能优于LMS算法，计算复杂度接近LMS算法。同时，结合RLS算法以及判决反馈变步长MOE(DF-MOE)算法，研究了RLS算法和DF-MOE算法相结合的盲多用户检测算法，该算法适应动态系统的能力增强，计算复杂度下降。 另外分析和仿真了基于Kalman滤波的盲多用户检测算法，并且综合分析比较了LMS算法、RLS算法和Kalman滤波的盲多用户检测算法的性能以及算法的计算复杂度。探讨了子空间盲多用户检测算法和Kalman滤波的盲多用户检测算法的结合算法，并进行了仿真，仿真结果表明该算法与Kalman滤波的盲多用户检测算法相比，收敛速度加快，稳态输出信干比相当。

CDMA has become one of the key techniques of 3G mobile communication system. However,multiple access interference(MAI) affects the system performance.Multiuser detection(MUD)is the way to restrain MAI.It is also good for restraining near-far effect and multipath interference. Blind multiuser detector has become an important study point as it has no needs of training sequence and the prior knowledge of interferential users.This thesis puts focus on blind MUD algorithms based on the current MUD technologies. Study the DS-CDMA system model of blind MUD. Make analysis and simulations the basic principle of minimum output energy(MOE) Least Mean Square(LMS) and Recursive Least Squares(RLS) two blind multiuser detection algorithm in synchronous DS-CDMA system under white Gaussian noise (AWGN) channel.On the base of deep research LMS algorithm, an improved variable step size LMS algorithm has been designed, simulation results show that performance of this algorithm is superior to LMS algorithm and it’s calculation is almost same as LMS algorithm.The algorithm of RLS combined with decision-feedback variable tep-size MOE(DF-MOE) is proposed for dynamic environment.The result from simulation indicates that this algorithm has good suppression effect on dynamic strong MAI and near-far effect. Make analysis and simulatons on Signal Interference Rate in different channel environment of Kalman algorithm of blind MUD, and three blind adaptive multiuser detection algorithm are analyzed and compared,including blind adaptive LMS algorithm, RLS algorithm and Kalman algorithm. Based on subspace blind multiuser detection and Kalman algorithm, we get a improving type Kalman blind multiuser detection algorithm. The simulation of computer shows that is superior to Kalman algorithm in the convergence performance.