【作者】 唐洪；

【导师】 邱天爽；

【作者基本信息】 大连理工大学 ， 信号与信息处理， 2006， 博士

【摘要】 在工程上和自然界中常常遇到一些偏离高斯分布的信号和噪声,这些信号和噪声常常伴有显著的脉冲性。如果对这些信号和噪声仍然采用高斯分布来模拟,则信号处理算法性能的退化是不可避免的。为了从理论上解决这一问题,自从20世纪90年代以来,Alpha稳定分布以其优良的性能被引入作为这些脉冲性信号和噪声的分布模型,并在此基础上逐步形成广义正态信号处理的理论框架。 鉴于各种通信系统中通信信道噪声表现出显著的脉冲性特征,比如:无线信道、浅海水声信道等等。在研究通信信道噪声的广义正态分布特性后,从恒模波束形成、恒模信道均衡、恒模信号恢复、子空间跟踪、自适应阵列和时间延迟估计等方面研究改善通信系统在脉冲性噪声环境中的性能。本文主要包括以下几个方面的工作: (1)提出了广义恒模算法和广义恒模阵列。无线信道和浅海水声信道的噪声都表现出较强的脉冲性,常规的恒模算法和恒模阵列性能出现退化,甚至不能正常工作。在广义正态信号处理理论框架内提出了广义恒模算法,算法韧性得到增强,并且常规恒模算法还是广义恒模算法的一个特例。以两个正弦信号组成的模型分析了广义恒模算法的增益过程,得到结论:在Alpha稳定分布噪声环境中,广义恒模算法捕获信号的性能由初始权向量和阵列输入信号的相对功率唯一确定。一般地,广义恒模算法将在输出端锁定具有较大功率的信源,并抑制掉其它信源。研究了浅海水声信道噪声的广义正态特性,并把广义恒模算法应用到水声信道均衡中取得了较好结果。以广义恒模算法和信号对消器组建了广义恒模阵列,分析并求解了广义恒模阵列的稳定点。研究了级联广义恒模阵列的多信号恢复能力,以及信号和Alpha稳定分布噪声在级联阵列间的传递行为。该项研究在以同信道干扰为主要危害的蜂窝通信系统中有潜在的应用价值。 (2)提出了两种韧性子空间跟踪算法,以及提出了基于子空间的韧性自适应阵列对用户信号的跟踪算法。①对最小特征值和相应特征向量的跟踪。把最小特征向量和最小特征值与一个二次型函数的最小点对应起来,实现了对最小特征向量的韧性跟踪。②对信号子空间的跟踪。把信号子空间的求解过程看作一个无约束的优化问题,在广义正态信号处理的框架下提出了新型代价函数,得到了韧性的信号子空间跟踪FLOS-ST算法,并采用M估计进行了进一步简化,得到了M-ST算法。把子空间跟踪应用在波达方向估计上,展现出这些算法的韧性。其中以M-ST算法的性能最好,韧性最强。以子空间的研究为基础,研究了IS-54/136系统中韧性自适应阵列对用户信号的跟踪。从子空间的角度利用整个TDMA时隙内的字符估计期望信号的传输向量,进而在阵列输出端重构期望用户信号。由于使用了子空间并利用了全部时隙内的字符,使得算法在脉冲性噪声更多还原

【Abstract】 In practice, many signals and noises are not exactly Gauss distributed. Furthermore, spikes and impulsiveness often accompany these signals and noises. If the signals and noises are still modeled by Gaussian distribution, the signal processing algorithms will degrade inevitably. In order to solve the problem theoretically, Alpha-stable distribution was introduced to model the signals and noises at the end of twentieth century. Based on the contribution of many people in the last decade, a new theory named Theory of Generalized Normal-distribution Signal Processing (TGNSP) comes into being.The noises in communication channels have been proved to be impulsive. To improve the performance of communication systems, the noises are modeled as Alpha-stable distribution in this paper. In the frame of TGNSP, the problems of beamforming, equalization of communication channel, signal copy, subspace tracking, adaptive array and time delay estimation (TDE) are studied. Main researches and conclusions of this paper are listed as follows:(l)The noises in wireless channel and the shallow water channel are often strongly impulsive. The traditional Constant Modulus (CM) algorithm and the CM array degrade greatly;even they cannot work any more under such impulsive noise condition. In the frame of TGNSP, the Generalized CM (GCM) algorithm is proposed. The GCM algorithm has strong robustness compared to the traditional CM algorithm. It can capture the desired signal in impulsive noise which is modeled as Alpha-stable distribution in this paper. On the other hand, the traditional CM algorithm is a special case of the GCM algorithm. An analysis model consisted of two sinusoid signal is adopted to study the gain process of the GCM algorithm. An important conclusion is obtained: In Alpha-stable noise environment, the performance that one of the input signals is captured by the GCM algorithm is only determined by the initial weight vector and the relative power of the input signals. Generally speaking, the GCM algorithm always captures one of the input signals with max power and suppresses others. The noises in shallow water channel are examed. It is proved that the noise is non-gaussian distributed with great amount of impulsiveness. The GCM algorithm is applied to equalization of shallow water channel. It is shown that the GCM algorithm improves the performance of Bit Error Rate (BER). The GCM beamformer and the new signal canceller construct the GCM array. The steady-state properties of the GCM array at convergence are examined. And the closed-form expression of steady point of the GCM array is derived in Alpha-stable noise. The cascade GCM array is also constructed to perform multi-signal copy. Furthermore, the signal-capture and direction-finding capabilities of each stage are discussed.This research has the potential application in cell mobile communication system in which the cochannel signal is the main interference.(2)Tow robust Subspace Tracking (ST) algorithms and one robust adaptive array algorithm based on subspace are proposed. The first ST algorithm is to track the minimum eigenvalue and corresponding eigen vector, where the eigenvalue and eigen vector are taken as the only minimum of a quadratic function and solution. The second ST algorithm is tracking the signal subspace, where the tracking process is looked as optimization problem without constraint. A new cost function for ST in Alpha-stable noise is proposed and a robust ST algorithm based on FLOS, i.e.FLOS-ST, is developed. In order to reduce computation load, The FLOS-ST is simplified by M-estimation. Then, a more robust ST algorithm, i.e.M-ST, is obtained. Based on the research about ST, a robust adaptive array for IS-54/136 system is studied. A new technique to estimate propagation vector by exploiting the whole frame of the received signal in Alpha-stable noise is proposed. Then, the combining weight vector of adaptive array can be estimated more steadily and the desired signal can be recovered with lower BER.(3)A wireless location technique based on TDOA (Time Difference of Arrival) is proposed in this paper, and a new wireless location system is developed. Some critical problems, such as denoise algorithms, impact of center frequency conversion on TDOA, and TDOA algorithms are studied. And some problems in application are also considered. The wireless source location system is constructed by embedding the software into corresponding hardware system. From the outdoor test, it is proved that the location system has high location accuracy. It is believed the location equipment will be seen in market soon.更多还原