【作者】 程皓；

【导师】 郭伟；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2007， 博士

【摘要】 直接序列扩频(Direct-Sequence Spread-Spectrum以下简称DS／SS)通信在保密通信和个人移动通信中获得广泛应用的同时，由于其具有低截获概率(Low Probability of Interception，以下简称LPI)、高抗干扰能力的特点，在侦察对抗和民用无线电资源监管领域成为非常棘手的难题。如何在非协作条件下，对隐藏在噪声中的直扩信号进行行之有效的盲检测和盲估计，成了现阶段军事侦察领域和民用监管领域的当务之急。同时，DS／SS信号的估计方法，本身也是宽带微弱信号检测领域以及通信对抗领域的一个重要研究课题。近年，发展了诸多针对于DS／SS信号的估计方法。这些估计方法大都基于常规信号检测与估计理论，但DS／SS信号通常是宽带微弱信号，并不符合常规信号处理理论，因此当观察信号的信噪比较低时，这些估计方法的估计性能恶化，甚至得不到正确的结果。从当前的研究状况来看，由于DS／SS信号的上述特性，对其估计的效果远不及普通信号理想，研究进展缓慢，概念性的研究较多，具体的研究方法还不成熟。考虑在未知信号各种特征参数(载频、码速率、伪码周期等)和伪码(PN码)码型的情况下对DS／SS信号进行全盲估计的挑战性，有必要对DS／SS信号的盲估计做进一步的研究。本文既归纳总结了前人的理论，又提出或改进了一些方法，使之这些方法能够形成一套完整的、行之有效的对DS／SS信号的盲估计流程。从而能对军事领域的侦察、(抗)干扰，民用领域的频率监管等方面起到积极的作用。同时，本文也对直扩码分多址(Direct Sequence Code Division Multiple Access以下简称DS／CDMA)信号的盲估计做了研究。本文的主要贡献在于两个方面：1．初步解决了特定调制条件下的低信噪比条件下DS／SS信号特征参数(包括载频、码速率、码长)估计问题；2．初步解决了码型估计问题(在码分多址下，解决了用户之间的码型分离问题)。本文研究了DS／SS信号及DS／CDMA信号盲(非协作接收)条件下的关键技术，在如下几个方面取得了成果和创新：混合使用了多种针对低弱信噪比下的谱估计算法，并应用到直扩信号中去，形成了一套有效的，针对该类信号特征参数的联合估计算法。能准确估计出DS／SS和DS／CDMA信号的特征参数(包括载频，码速率，码长等)。这几种算法都能针对该类信号低信噪比的特点，在低信噪比条件下有效工作，并且互为补充，形成了一套完整的参数估计流程。重新构造出一种用于矩阵分解的自相关矩阵。通过对新构造的自相关矩阵的特征值分解，能有效地克服传统子空间方法对噪声敏感的缺点，在低信噪比情况下，能有效解决信号特征值被噪声特征值淹没(分辨率不足)的问题。这种方法用于直序列扩频信号的扩频码序列估计中，仿真效果与理论计算一致。同时，该方法也能估计出由随意采样点不同引起的失步时间，并能估计出系统中存在的信噪比。受神经网络能自主学习这一思想的启发，设计了一种用于DS／SS信号PN码序列估计的神经网络方法。在PN码序列估计的矩阵分解得到失步时间的基础上，提出了用于信号PN码估计的无监督自组织映射神经网络算法，并且验证了它的有效性。设计出一种利用延时相乘法DS／CDMA用户数目与用户参数的联合估计法。针对实际的CDMA通信网，由于各移动台位置和传输环境的不同，反向信道中的用户信号通常是异步的。对于基站而言，它所收到的DS／CDMA混合信号是由多个初相不同的DS／SS信号组成。同时，各用户DS／SS信号的幅值也可能不同。在这种情况下，可以通过延迟相乘梳状滤波的方法，利用各用户信号的不同初相，在时域上分辨出用户的个数，同时估计出各自的相位和幅值。提出利用相位信息重复，用于多用户直序列扩频码分多址(DS／CDMA)系统的特征码快速盲分离算法。该算法利用DS／CDMA系统独有的多用户信息混合后相位重复特性，在已估计出其它参数前提下，采用基带接收，同相相加、反相相减、无关项舍弃的简单操作，实现多用户特征序列的快速分离。该方法较ICA算法有更少的计算量，保留了信道内多用户的幅值特征。不存在ICA方法中若混合信号同时存在两个以上高斯信号，或同时存在超高斯信号的欠高斯信号的情况下，算法失效的问题。改进了一种在瞬时混合分离条件下性能优良的盲分离(BSS)算法，在准则不变条件下，改进成为针对于多通道盲解卷(MBD)的算法。并将这一算法应用于非平稳态的DS／CDMA信号分离中，验证了该算法的有效性。上述信号估计方法都是适用于盲条件下的，无需知道观察信号的特征参数与PN码序列，且都能工作在较低的信噪比条件下。而且上述的估计方法优缺点互为补充。仿真实验表明，总体上这些方法可以作到较低的信噪比容限。更多还原

【Abstract】 Direct sequence spread spectrum (DS/SS) signals have been widely used in secure communication and mobile communication known as Code Division Multiple Access (CDMA) system for the reason is that DS/SS signals have many advantages such as anti-jamming capability, low probability of interception and multiple access capability. The interception and radio monitoring for spread spectrum communication have been of great research interests. The problem of estimating DS/SS & DS/CDMA signals direct from the received signals has been of great research interest with the development of the field of wide-band weak signal processing and the field of communication antagonism. The dissertation focus on detecting and estimating the DS/SS and direct sequence code division multiple access (DS/CDMA) signals hidden in the noise.For the last several decades, many methods for the estimation the DS/SS & DS/CDMA signal estimation methods has been developed. Most of these methods are based on the theory of conventional signal detection and estimation. The DS/SS &DS/CDMA signals are wide-band weak signals, they are not in conformity with the theory of conventional signal processing, when the signal to noise ratios of the received signals becomes lower, the performance of the methods take a turn for the worse. Though DS/SS &DS/CDMA signals are widely used in the field of military and civil, recent researches on the DS/SS &DS/CDMA signal estimation appeared in the literature. The development of DS/SS &DS/CDMA signal estimation is very slow, it is almost in study of conception and the research on concrete method is far from comprehensive and mature. Due to the challenge of DS/SS &DS/CDMA signal estimation without any prior knowledge of the pseudo-noise (PN) sequence, the study of DS/SS &DS/CDMA signal estimation is necessary.The main contributions of this dissertation include two aspects. One is parameter estimation of the DS/SS &DS/CDMA signal, include period estimation and chip interval estimation of the PN sequence etc., the other one is PN sequence estimation of the DS&DS/CDMA signals.Several valuable and important results which bring forth new ideas are achieved and listed as follows:A union approach based on the various spectrum estimation algorithms has been presented. Through the new approach, we can estimate some characteristic parameter, including carrier frequency, code length and chip rate of the DS and DS/CDMA signal. Besides, these methods can make up for their defects each other.In order to solve the estimation problem of PN sequence itself, a matrix decomposition approach to estimate the PN sequence itself has been presented. Through decomposition of the correlation matrix of DS&DS/CDMA signal, we can estimate the PN sequence from the largest and the second largest left singular vector. The final correlation matrix R, used in Eigen value decomposition (EVD) or singular value decomposition (SVD), can be denoted as the received vector multiplies itself transposition. A new method of constructing the correlation matrix is proposed in this dissertation. The proposed algorithm can enhance the signal Eigen values’ resolving power and resolve the problem that traditional subspace methods can not be applied in low SNR. Then the proposed new method is applied to the direct sequence spread spectrum (DS/SS) signal’s signature sequence estimation. Its performance is analyzed, and some illustrative simulations are presented.Having not the apriority knowledge about the DS/SS signal in the non-cooperation condition, we apply self-organizing feature map (SOFM) neural network theory to detect and identify the signal parameter and PN sequence. The computer simulation and experiment test have demonstrated that the algorithm is effective. Comparing the traditional slip-correlation method, the BER of SOFM algorithm and implementation complexity is lower.An approach for solve the estimation of users number has been presented. This algorithm is based on the delay-multiply. According to the different phase and amplify of each mobile phone and the different transmission condition, each user can be separated in the time field.A blind separation algorithm for instantaneous mixed DS/CDMA signal with unknown spreading codes is discussed in this dissertation. The reappearance of mixing users’ phase information is utilized to separate the users’ sequence by the simply plus and subtract operation in low SNR. The proposed method is different for ICA that not only saved the users’ amplitude information but also calculation complication is independent of chip length L. The implement condition of the algorithm is analyzed and the performance of parameter estimation is measured by computer simulation.An improved and more complex BSS algorithm for separating linear convolved mixtures of no stationary signals in CDMA system is presented. This algorithm relies on the no-stationary nature of the sources to achieve separation. Most of which assume statistically stationary sources as well as instantaneous mixtures of signals. In practicality, the CDMA sources received are no-stationary and linear convolute mixing. A more complex BSS algorithm is required to achieve better source separation. The algorithm is based on minimizing the "average squared cross output channel correlation". The mixture coefficients are totally unknown, while some knowledge about temporal model exists. The simulation results show the effectiveness of the method in the blind detection of DS/CDMA signals.This methods above mentioned are suited for blind condition that is all the parameter not been known. Through the computer simulation, all of these algorithms can work in lower SNR.更多还原