多维参数估计的快速算法研究

发布时间：2018-04-06 01:02

本文选题：二维DOA快速算法　切入点：L阵　出处：《电子科技大学》2014年硕士论文

【摘要】：利用平面阵列估计多信号的二维DOA(Direction-of-Arrival)参数在许多军事及国民经济领域具有重要作用,比如声呐,雷达和通信等。由于二维DOA维数的增加,DOA估计过程的计算量主要受阵列结构以及对估计的俯仰角和方位角配对的影响,然而传统的配对算法主要是通过二维搜索和求解非线性优化问题得到的,计算比较复杂。因此本文围绕算法的计算量等问题,主要研究基于L阵的二维DOA参数估计算法,重点研究了以下几个方面的内容:1、研究了阵列的接收模型,以及各种平面阵列天线的延迟和相位计算的区别;同时简单介绍了MUSIC(Multiple Signal Classification)算法、ESPRIT(Estimate Signal Parameters via Rotational Invariant Techniques)算法、PM(Propagator Method)算法和MSWF(Multi-Stage Wiener Filter)算法,对上述1-D DOA估计算法进行对比仿真,分析了上述算法计算复杂度和角度估计的性能,同时指出了在1-D DOA估计算法中减少算法计算量的实现方法。2、研究了利用L阵来估计窄带信号二维到达角的算法,包括利用自相关矩阵的修正的PM和CCM-ESPRIT(Cross-Correlation Matrix based on ESPRIT)算法,以及利用互相关矩阵的JSVD(Joint Singular Value Decomposition)和CODE(Computationally efficient cross-correlation based 2-D DOA Estimation)算法,对上述算法的由来以及性能给出了解释并分析每个算法的计算复杂度,同时关注了在算法中存在的问题;最后提出了一种利用MSWF快速估计特征子空间的算法,该算法针对互相关矩阵进行角度估计,无需谱峰搜索和SVD,计算复杂度较低,同时算法可自行配对,算法的实时性较好,有利于工程应用。3、针对宽带线性调频信号(Linear Frequency Modulation,LFM)的多维参数估计,以及LFM信号在分数阶傅里叶变换(Fractional Fourier Transform,FRFT)的特性,使用分数阶傅里叶变换估计出初始频率和调频斜率;同时在LFM信号的解线调FRFT域上提出了利用PM和MSWF估计DOA的快速算法,上述算法无需EVD(Eigen-Value Decomposition)或SVD,虽然估计角度精度有所损失,但算法的计算复杂度低于基于EVD的估计算法。最后本文通过上述研究和仿真实验得到一些很有参考意义的结论,对阵列信号处理中的多维参数估计的快速算法有一定的借鉴意义。
[Abstract]:Using planar array to estimate multi-signal 2-D DOA Direction-of-Arrival (DOA) parameters plays an important role in many military and national economic fields such as sonar radar and communication.Due to the increase of two-dimensional DOA dimension, the computational complexity of the estimation process is mainly affected by the array structure and the pairing of pitch and azimuth angles. However, the traditional pairing algorithm is mainly obtained by two-dimensional searching and solving nonlinear optimization problems.The calculation is more complicated.Therefore, this paper mainly studies the algorithm of 2-D DOA parameter estimation based on L-matrix, focusing on the following aspects: 1, the receiving model of the array.At the same time, the paper introduces the MUSIC(Multiple Signal classification algorithm, the estimation Signal Parameters via Rotational Invariant Techniques-Esprit algorithm and the MSWF(Multi-Stage Wiener filter algorithm, and compares and simulates the 1-D DOA estimation algorithm mentioned above, and introduces the difference between the delay and phase calculation of various planar array antennas, and gives a brief introduction to the MUSIC(Multiple Signal classification algorithm, the estimation Signal Parameters via Rotational Invariant Techniques-algorithm, and the MSWF(Multi-Stage Wiener filter algorithm, and compares and simulates the above 1-D DOA estimation algorithm.The computational complexity and angle estimation performance of the above algorithms are analyzed. At the same time, the realization method of reducing the computational load in the 1-D DOA estimation algorithm is pointed out, and the algorithm of estimating the two-dimensional arrival angle of narrowband signals by using L matrix is studied.It includes modified PM and CCM-ESPRIT(Cross-Correlation Matrix based on Esprit algorithm using autocorrelation matrix, JSVD(Joint Singular Value decompositiontion using cross-correlation matrix and CODE(Computationally efficient cross-correlation based 2-D DOA estimation algorithm.The origin and performance of these algorithms are explained and analyzed, and the computational complexity of each algorithm is analyzed. At last, an algorithm using MSWF to estimate feature subspace is proposed.Based on the angle estimation of cross-correlation matrix, the algorithm does not need spectral peak search and SVD, and its computational complexity is low. Meanwhile, the algorithm can be matched by itself, and the algorithm has better real-time performance.In view of the multidimensional parameter estimation of linear Frequency modulation (LFM) for wideband LFM signals and the characteristics of LFM signals in fractional Fourier transform (Fractional Fourier transform), the initial frequency and frequency modulation slope are estimated by fractional Fourier transform.At the same time, a fast DOA estimation algorithm based on PM and MSWF is proposed in the FRFT domain of LFM signal unalignment. These algorithms do not need EVD(Eigen-Value decomposition.Although the estimation angle accuracy is lost, the computational complexity of the algorithm is lower than that of the EVD based estimation algorithm.Finally, through the above research and simulation experiments, some useful conclusions are obtained, which can be used for reference in the fast algorithm of multi-dimension parameter estimation in array signal processing.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.23

【参考文献】