节点文献

空时联合多用户检测技术的研究

The Technique of Space-Time Multi-user Detection

【作者】 王香玉

【导师】 王珂;

【作者基本信息】 吉林大学 , 信号与信息处理, 2004, 硕士

【摘要】 多用户检测技术是 CDMA 系统中最有效的抗多址干扰技术,是一种从接收端的设计入手的干扰抑制方法。它在传统检测技术的基础上,充分利用干扰用户的信息(扩频序列相关特性,信号幅度变化,信号同步特征等)来消除或减轻多址干扰的影响,以达到可靠地提取有用信号。而空时处理是借助阵列天线技术利用信号的空间特征来抑制干扰。将两者结合起来组成空时联合的多用户检测器是近几年来研究的热点。 与 TDMA,FDMA 相比,CDMA 具有系统容量大、抗干扰能力强、频谱利用率高、发射功率低、保密性能好等优越性,这使得 CDMA 多址技术成为IMT-2000 研究和开发的技术基础。但是所有的 CDMA 用户占用同一时间和同一频带,接收信号是所有用户信号的总和,由于异步和多径传播等原因,接收信号中各个用户的信号存在一定的相关性,使得接收端只借助每个用户的扩频码进行解扩时,非期望用户的信号对期望用户信号的正确解调造成干扰,这种干扰称为多址干扰(MAI)。由个别用户产生的 MAI 虽然很小,随着用户数的增加或信号功率的增大,MAI 成为宽带 CDMA 通信系统的一个主要干扰。MAI 的存在影响了通信的质量,限制了系统容量,从而使得各种抗多址干扰(MAI)技术的研究有了重要的意义。 基于维特比算法的最优空时多用户检测算法的复杂度随着用户数的增加而呈指数增长的趋势,无法实时实现,因此次优的空时多用户检测的研究成为了人们研究的重点。次优的线性时空多用户检测计算复杂度大大降低,性能接近最优的空时多用户检测,它分为最小均方误差(MMSE)空时多用户检测和解相关空时多用户检测。然而它们目前存在的问题是:解相关空时多用户检测涉及矩阵求逆,运算量较大,并且解相关操作使背景噪声增强;MMSE空时多用户检测在消除 MAI 和不增强背景噪声之间进行了折衷,但是它需要估计信号幅度,抗远近效应方面性能不如解相关空时多用户检测,并且运算量和解相关算法相近。 针对以上问题,我们做了以下研究: 63<WP=70>吉林大学硕士学位论文 1. 研究了线性空时多用户检测中的解相关空时多用户检测技术,提出了一种针对基站接收的去偏解相关空时多用户检测器,其计算量较小,性能优于一般的解相关空时多用户检测器。文中采用的是直接序列扩频 CDMA 系统(DS-CDMA),信道是多径 CDMA 信道,假设每一个符号的多径传播都被限制在一个符号间隔内。通过分析知,每个符号的信息都包含在以此符号周期为中心的相邻的三个符号周期中,即符号间干扰(ISI)只来自于相邻三个比特。去偏的解相关空时多用户检测器取相邻三个信息比特的充分统计量(包含了原观测值与最佳决策有关的所有信息的统计量),对其进行解相关操作后,它考虑了偏差项的影响而只取中间的信息比特作为估计;如果把这三个信息比特全部作为估计而不考虑偏差项的影响得到的是有偏的空时解相关多用户检测。同时处理大于三个信息比特的充分统计量可以得到更多信息比特的信息,但这是以增加计算复杂度为代价的,而在文中假设下三个比特的信息已经足够。另外,解相关操作涉及对矩阵求逆,此矩阵是一个对角占优矩阵,其中含有大量的零元素,所以适合采用高斯-塞德尔迭代算法,这种算法有很好的收敛性能。我们做的仿真实验如下: 1) 对系统中两个信号比较弱的用户分别采用如下三种检测器:有偏的空 时解相关多用户检测器、去偏的空时解相关多用户检测器和空时匹配滤 波器,记录他们的比特误码率随信噪比变化的情况,并给出了变化曲线。 取信号弱的用户是因为他们更能体现检测器的性能。 通过对仿真结果的分析比较我们得出结论:去偏的空时解相关多用户检测器的性能优于其它两种检测器。 2) 记录每个用户每一次迭代的比特误码率随信噪比变化的情况,此实验 用来表明高斯-塞德尔迭代算法的收敛性。从图中可以看出,这种算法在 4-5 次迭代内就收敛了,并且也表明了最大的性能改善表现在第二次迭代, 从而说明这种迭代算法具有较好的收敛性能。 上述都是假设所有用户特征序列已知的检测,不能消除其它小区的 MAI对本小区的影响,也不能用于下行链路的接收而只适合于基站接收的情况,于是人们开始研究所有干扰用户的特征序列未知的盲空时多用户检测,它适合于移动台的接收。在异步或多径传播的情况下,不同用户的信息符号间不同步,因此在被接收用户信息符号的持续期内外,干扰用户和被接收用户的 64<WP=71>摘要信号间是相关的。为了最大限度地抑制干扰,考虑持续期外的信号是有意义的,所以数据窗口宽度的选取也就变得非常重要。因此,本文还做了如下研究:2. 研究了基于 LMS 算法的盲空时多用户检测器,并给出其组成结构图,重点分析四种数据窗口选择方案对接收数据中包含的多址干扰和符号间干扰的影响,这四种方案是:(1)捕获了期望用户所有的路径上的信号,充分利用所有多径上的信号能量;(2)充分利用延迟最短的路径的信号的能量及其相应的其它路径的部分信号能量;(3)充分利用延迟最长的路径的?

【Abstract】 Multi-User Detection(MUD) technique is the most effective anti-MAI techniquein CDMA system. It is a method of anti-interference based on the receiver’s design.It can demodulate the desired user’s signal reliably by making full use of otherusers’ information. While space-time processing reduces the interference in virtueof the signal’s spacial property obtained from the antenna array. Combining thetwo techniques into space-time MUD (ST-MUD) is heatedly discussed in recentstudies. Compared with FDMA and TDMA, CDMA is superior in high capacity,anti-interference, frequency application, low transmitting power and high privacy,etc., which makes CDMA become the technical foundation of IMT-2000. However,in CDMA system, all users share the same time and frequency band and thereceiving signal is the sum of all users’ signal. Different users’ signal is correlated,because of asynchronous and multipath spreading of wireless signal. Although theMAI produced by individual user is small, it will become the primary interference,with the increase of the users’ number and the signal’s power. The existence ofMAI limits the capacity of CDMA, and affects other aspects of its superiority.Hence the study of various anti-MAI techniques has great significance. Optimum ST-MUD based on Viterbi algorithm can’t be implemented in realtime because its computational complexity increases in exponent with the increaseof the users’ number, which restricts its application and some sub-optimumST-MUDs appear. MMSE ST-MUD and Decorrelation ST-MUD are two typicalsub-optimum linear ST-MUDs. Their computational complexity lessens greatlywhile their performances are close to optimum ST-MUD’s. However, decorrelationST-MUD involves computing the inverse of a matrix, and the operation amount isrelatively great. In addition, the operation of decorrelation makes the ambientnoises strong. MMSE ST-MUD compromises between eliminating MAI andbuilding up no more ambient noises, but it needs to estimate the signal amplitude, 66<WP=73>ABSTRACTand the performance of “near-far” resistance is not so good as DecorrelatingST-MUD and the computational complexity is close to it. Take the above problems into consideration, the following researches are carriedout: 1. Studying the decorrelating ST-MUD technique and proposed a de-biasingdecorrelating ST-MUD, whose computational complexity is comparatively smallwhile performances are superior to the ordinary decorrelating ST-MUD. In thispaper, DS-CDMA (Direct Sequence Spreading CDMA) system is adopted. Signalspreads in multipath channel and it is assumed the multipath delay of each symbolis confined in one symbol’s period. Under this assumption, the information of eachsymbol is in the period of three continuous symbol centered on it, i.e. ISI onlycomes from these three bits. The de-biasing decorrelating ST-MUD takes theefficient statistics of three continuous symbol and decorrelating them, and thentakes the middle bit as the estimation. If we take these three bits as the estimationand don’t take the effect of error into account, a biasing decorrelating ST-MUD isobtained. In addition, the burden of work out the inverse of a matrix is heavy, sowe adopted the Gauss-Seidel iteration algorithm which is suitable for the matrix inthis paper and is superior in the convergence. Therefore we obtained a de-biasingdecorrelating ST-MUD which is simple-structured and low computationalcomplexity. The emulation experiments which have been done are as follows: 1) Comparing the BER (bit error rate) of each user when he adopt the followingthree detectors relatively: biasing decorrelating ST-MUD, de-biasing decorrelatingST-MUD and space-time matched filter. In the experiment two low-powered usersare recorded because they can test the detectors’ performances more efficiently.Through analyzing of the results of the experimental we draw the conclusion thatthe de-biasing decorrelating ST-M

  • 【网络出版投稿人】 吉林大学
  • 【网络出版年期】2004年 04期
  • 【分类号】TN929.533
  • 【被引频次】1
  • 【下载频次】291
节点文献中: