基于独立分量分析的DS-CDMA多用户检测

【作者】 吴蓓；

【导师】 李晓辉；

【作者基本信息】 安徽大学 ， 电路与系统， 2003， 硕士

【摘要】 码分多址接入(CDMA:Code Division Multiple Access)方式以频谱利用率高、抗多径、抗干扰、软容量、低功率、软切换、宏分集、频率规划简单、用户接入方便等其它多址技术不可比拟的优越性被认为是未来移动通信系统的理想接入技术之一。在CDMA系统中各用户是根据地址扩频码的不同，即根据它们之间互相关特性来区分的。若用户间的互相关不为零，则会导致多址干扰的产生；另一方面，由于CDMA系统是干扰受限的系统，即干扰程度直接影响到系统软容量和系统内用户服务质量的高低，为了有效地克服多址干扰的影响，多用户检测(MUD:Multi-user detection)技术的使用就显得尤其重要。传统的CDMA检测器分别对每个用户进行扩频码的相关运算，这种做法没有考虑多址干扰(MAI:Multiple Access Interference)和普通白噪声的差异。MUD则充分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，从总体上提高各个用户的性能。它克服了多址干扰，更加有效地利用上行链路频谱资源，显著增加系统容量。 本文比较全面的从第二章到第四章分别介绍了CDMA扩频通信的原理和现有的各种多用户检测技术以及独立分量分析(ICA:Independent Component Analysis)这种盲源分离算法，并在此基础上，从多用户信号的统计独立同分布的统计特征出发，分析研究了CDMA通信中多用户信号的数学模型，将盲源分离技术应用到多用户信号检测当中来，提出了两种盲多用户检测的策略：基于独立分量分析的下行链路延迟估计的多用户检测算法；基于独立分量分析后处理的匹配滤波多用户检测。 本文在第五章详提出了这两种盲多用户检测策略。传统的信道估计算法需要训练序列使接收端的参数调整到理想状态，而本文提出的基于ICA的信道估计的多用户检测算法不需要训练序列，它是利用下行信道的固有特点，用ICA的盲源分离法估计出多径信道的卷积矩阵，从而从中提取出信道的延迟信息，仿真实验结果证明这种方法在节省了频谱资源的同时取得较好的估计效果，使得传统的接收机的误码性能得到了很大的提高。基于ICA后处理的匹配滤波多用户检测是用传统的检测器的输出来初始化ICA的迭代，它不但充分利用多用户信号的已知信息，克服了ICA的不确定性问题，同时也充分利用了多用户信号的统计独立性，仿真实验结果证明这种多用户检测算法在高信噪的情况下，误码性能改善随着信噪比的提高不断增加。更多还原

【Abstract】 CDMA (Code Division Multiple Access) is a promising technology for future mobile communication system because of its several features with which other multiple access technology can not compare: higher efficiency of frequency allocation, robust anti-multipath effect, macro diversity, mitigating interference, soft capacity, low power, soft handoff, simple frequency design, easy access, and so on. In CDMA communication system, the users are identified by their different spread spectrum codes which are unique to each user, namely they are identified by the cross correlation of spread spectrum codes. If the correlation of different spread spectrum codes is not zero, there would be multi-access interference. Additionally, CDMA communication system is a interference-limited system, that is to say that the interference directly influences the capacity of this system and the QoS of the user in the system. In order to rid the multi-access interference, multi-user detection (MUD) plays a important role. Traditional CDMA detector executes the correlation operation on the receivers signal to get the user’ symbol, but this method does not considered the difference between the multi-user interference and the white noise. Multi-user detection executes the jointly detection fully using the spreading sequence, time delay x phase information and amplitude, the performance has enhanced as a whole. It overcomes the multi-user interference, the uplink frequency resource is used efficiently, so the capacity of system has been increased notably.In this paper, the principle of spread sequence communication and all kind of the current multi-user detection techniques and the independent component analysis (ICA) algorithm are respectively analyzed completely in chapter II , chapter III and chapter IV. Based on the statistical feature of multi-user signal that the signal of every user’s and every channel is statistical independent with each other and take the same distribution, the mathematical model of the received multi-user signal is fully researched, then the blind signal detection algorithm is introduced into the multi-user signal detection, two new algorithms of multi-user detection are proposed: multi-user detection based on the delay estimation of downlink using ICA in DS-CDMA and based on ICA post processing matched filter multi-user detection.In the fifth chapter, these two algorithms are recommended in detail. Traditional delay estimation based on ICA requires the trail sequences to initialize the receiver, but the new algorithm based on ICA does not need the trail sequences. It is based on the channel character of downlink,using the ICA algorithm to estimate the multi-path mixture matrix, then, find the delay information which is embodied by the column vector of the mixture matrix. The simulation results show that it does enhance the performance of traditional detector without wasting the invaluable frequency resource. Based on ICA post processing matched filter multi-user detection lets the output of traditional matched filter to initialize the ICA iterations, not only the known spread information of interesting user is used to overcome the uncertainness of ICA, but also the character of statistical independence is used. The simulation results show that it advances the ability of traditional detector when the signal-noise-ratio is large.Keywords Code Division Multiple Access; Independent Component Analysis; Channel Estimation; Matched Filter; Blind Multi-user Detection更多还原