【作者】 娄学伟；

【导师】 覃亚丽；

【作者基本信息】 浙江工业大学 ， 信号与信息处理， 2010， 硕士

【摘要】 波达方向(DOA)估计是阵列信号处理领域的一个重要研究方向,应用非常广泛,非平稳宽带线性调频(LFM)信号的DOA估计是波达方向(DOA)估计一个新的研究热点。作为一种新的时频分析工具,分数阶Fourier变换可以理解为一种线性的时频处理方法,用来处理LFM类信号时具有一些其他的分析工具所没有的优点。结合阵列信号处理技术,本文把FRFT应用到对LFM信号的DOA估计领域,并对其进行了相应地探讨。1、论文首先对分数阶Fourier变换的定义进行了详细地介绍,分析了一些基本的性,并给出一种离散算法,在此基础上分析了高斯白噪声在分数阶Fourier变换域的性质。2、分析了DOA估计技术的一些经典算法,在建立信号的接收模型,针对窄带不相干平稳信号源,对延迟-相加法和Capon最小方差法进行简单的描述,并做了简单的对比,Capon最小方差法的分辨能力要比对延迟-相加法有所改善;接着对多重信号分类算法(MUSIC)算法和信号子空间特征矢量生成广义特征值(GEESE)算法进行了详细地分析。针对对窄带不相干平稳正弦信号源,这两种方法可以有效的对信号进行DOA估计,与延迟-相加法和Capon最小方差法相比,这两种算法的估计性能要优越的多,并分析了阵元数目对估计分辨能力的影响。3、从LFM信号的时频特性出发,对其在分数阶Fourier域的特性进行了分析。利用LFM信号在特定的FRFT域呈现能量聚集的特性,在相应的分数阶Fourier域,根据天线阵列各阵元接收信号之间的相位差关系,构造出信号在分数阶Fourier域的时不变方向向量。然后分别采用MUSIC算法和GEESE算法对宽带LFM信号进行DOA估计。从而实现LFM信号的DOA估计,从算法原理、实现过程等方面,分析了该算法,并进行了仿真分析。更多还原

【Abstract】 Direction of Arrival (DOA) estimation has been a significant research area in array Signal processing. The DOA estimation of the non-stationary Linear Frequency Modulation (LFM) signal has been becoming a new hot topic and it has been received growing attention in recent years. With good cross-terms reduction and high time-frequency resolution, the Fractional Fourier Transform (FRFT) has been considered a linear and full time domain analysis tool for non-stationary signal processing. For a newly developed time-frequency analysis tool,it has been widely used in the multi-component LFM signal processing for its perfect property. Combining the FRFT with array Signal processing,the DOA estimation of LFM signal is discussed based on the FRFT in this thesis.1、We presented the definition of FRFT and some basic characteristics, and analysis a discrete algorithm. Then the energy distribution of Gaussian white noise in the FRFT domain also is discussed.2、Analyze some classic algorithm for the DOA estimation. We establish signal reception model firstly, then the Delayed-add and Minimum Variance Distorionless Response (MVDER) estimation algorithm have been discussed for the narrowband non-coherent stationary signal source, compare this two algorithm we can get that the resolving power of the MVDER algorithm’s is better than the Delayed-add algorithm. Further more, anther two methods for DOA estimation are discussed: Multiple Signal Classification (MUSIC) algorithm and Generalized Eigenvalues utilizing Signal subspace Eigenveetors (GEESE) algorithm. In the same condition, the resolving power of MUSIC and GEESE algorithms are better than Delayed-add and MVDER algorithm. But all these algorithms can not estimate the DOA when signal is non-stationary LFM signal. We also discussed the influence for MUSIC algorithm performance when the number of array is different. Simulation results proved that3、According to the time-frequency properties of LFM signal, we have analyzed some properties of LFM signal in the FRFT domain, consider the LFM signal in certain fractional Fourier domain can take on energy concentration property, and the phase relationship between the received signal of received array element, the time-invariant direction matrix in that FRFT domain is derived,then the signal direction can be estimated by the MUSIC and GEESE algorithm, GEESE algorithm has a low complexity. The simulation results show that the algorithm can effectively estimate the signal direction even in a low Signal to Noise Ratio (SNR).更多还原