基于均匀圆阵的MUSIC快速测向算法研究
发布时间：2018-07-28 14:05
【摘要】：基于均匀圆阵的MUSIC测向算法具有优秀的性能,但是涉及的运算复杂度很高,难以满足实时性的要求。本文主要研究均匀圆阵的MUSIC快速测向算法,具体如下:(1) UCA-RB-MUSIC算法首先采用波束空间变换将均匀圆阵复杂的阵列流型变换为类似均匀线阵的范德蒙结构,然后采用一种酉变换实现波束空间数据协方差矩阵由复数域到实数域的转换。本文针对UCA-RB-MUSIC算法中酉变换运算复杂的问题,提出一种新的酉变换矩阵,其在不损失性能的同时减少了运算复杂度。(2)子空间估计在高分辨率测向算法中占有举足轻重的地位,但是它一般需要计算阵元接收数据协方差矩阵并对其做特征分解,运算复杂度很高。对此,本文首先研究了特征分解的硬件实现算法,并对单侧Jacobi算法进行改进,使其在迭代运算的同时以很小的硬件开销实现特征向量按特征值大小的自动排序。其次,本文研究了基于多级维纳滤波(MSWF)的快速子空间估计技术,并对其在低信噪比情况下性能急剧下降的问题给出一种改进方法。该方法改进了低信噪比情况下的MSWF类子空间估计性能,但同时有一定概率引起伪峰的出现。(3)在研究波达方向估计问题过程中一般假设信源数为己知的,然而其在实际应用中通常是需要估计的。对此,本文研究了基于MSWF的信源数估计方法,并针对其在低信噪比情况下性能急剧下降的问题提出一种改进方法。(4)本文针对均匀圆阵的求根MUSIC算法中子空间估计过程运算复杂、耗时大的问题,引入基于MSWF的快速子空间估计技术,同时对其递推过程采用一种预处理方法,使运算过程由复数运算转为高效的实数运算。均匀圆阵中MSWF类子空间估计方法依然存在低信噪比情况下性能急剧下降的问题,对此本文提出另一种改进方法。该方法针对性能下降的本质原因,大幅削弱噪声对子空间估计性能的影响,但是该方法需要预估噪声功率。MSWF类子空间估计方法与均匀圆阵求根MUSIC算法有多种组合方式,文章最后通过仿真实验分析了多种组合方式的测向性能。
[Abstract]:The MUSIC direction finding algorithm based on uniform circular array has excellent performance, but the computational complexity involved is very high, so it is difficult to meet the requirement of real-time. In this paper, the MUSIC fast direction finding algorithm for uniform circular array is studied. The main results are as follows: (1) the complex array flow pattern of uniform circular array is transformed by UCA-RB-MUSIC algorithm into a Vandermonde structure similar to uniform linear array. Then a unitary transformation is used to transform the covariance matrix of beamspace data from complex domain to real field. Aiming at the complexity of unitary transformation in UCA-RB-MUSIC algorithm, a new unitary transformation matrix is proposed in this paper, which reduces the computational complexity without loss of performance. (2) subspace estimation plays an important role in high-resolution direction-finding algorithm. However, it usually needs to compute and decompose the covariance matrix of data received by elements, so it has a high computational complexity. In this paper, the hardware implementation algorithm of feature decomposition is studied firstly, and the one-sided Jacobi algorithm is improved to realize the automatic sorting of eigenvector by eigenvalue at the same time of iterating operation with very small hardware overhead. Secondly, the fast subspace estimation technique based on multistage Wiener filter (MSWF) is studied, and an improved method is proposed to solve the problem that its performance drops sharply in the case of low signal-to-noise ratio (SNR). This method improves the performance of MSWF subspace estimation under low SNR, but at the same time there is a certain probability to cause pseudo peaks. (3) in the study of DOA estimation, it is generally assumed that the number of sources is known. However, it usually needs to be estimated in practical applications. In this paper, the source number estimation method based on MSWF is studied, and an improved method is proposed to solve the problem of the sharp deterioration of its performance in the case of low signal-to-noise ratio. (4) the neutron space estimation process of MUSIC algorithm for uniform circular array is complex. In this paper, a fast subspace estimation technique based on MSWF is introduced, and a preprocessing method is used in the recursive process to change the operation from complex operation to efficient real number operation. The MSWF subspace estimation method in uniform circular array still has the problem of sharp degradation of performance in the case of low signal-to-noise ratio (SNR). In this paper, another improved method is proposed. The effect of noise on the performance of subspace estimation is greatly weakened for the essential cause of performance degradation. However, this method needs to estimate the noise power. MSWF-class subspace estimation method and uniform circular array root-finding MUSIC algorithm have many combinations. Finally, the performance of multiple combination methods is analyzed by simulation experiments.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.7
本文编号：2150447
[Abstract]:The MUSIC direction finding algorithm based on uniform circular array has excellent performance, but the computational complexity involved is very high, so it is difficult to meet the requirement of real-time. In this paper, the MUSIC fast direction finding algorithm for uniform circular array is studied. The main results are as follows: (1) the complex array flow pattern of uniform circular array is transformed by UCA-RB-MUSIC algorithm into a Vandermonde structure similar to uniform linear array. Then a unitary transformation is used to transform the covariance matrix of beamspace data from complex domain to real field. Aiming at the complexity of unitary transformation in UCA-RB-MUSIC algorithm, a new unitary transformation matrix is proposed in this paper, which reduces the computational complexity without loss of performance. (2) subspace estimation plays an important role in high-resolution direction-finding algorithm. However, it usually needs to compute and decompose the covariance matrix of data received by elements, so it has a high computational complexity. In this paper, the hardware implementation algorithm of feature decomposition is studied firstly, and the one-sided Jacobi algorithm is improved to realize the automatic sorting of eigenvector by eigenvalue at the same time of iterating operation with very small hardware overhead. Secondly, the fast subspace estimation technique based on multistage Wiener filter (MSWF) is studied, and an improved method is proposed to solve the problem that its performance drops sharply in the case of low signal-to-noise ratio (SNR). This method improves the performance of MSWF subspace estimation under low SNR, but at the same time there is a certain probability to cause pseudo peaks. (3) in the study of DOA estimation, it is generally assumed that the number of sources is known. However, it usually needs to be estimated in practical applications. In this paper, the source number estimation method based on MSWF is studied, and an improved method is proposed to solve the problem of the sharp deterioration of its performance in the case of low signal-to-noise ratio. (4) the neutron space estimation process of MUSIC algorithm for uniform circular array is complex. In this paper, a fast subspace estimation technique based on MSWF is introduced, and a preprocessing method is used in the recursive process to change the operation from complex operation to efficient real number operation. The MSWF subspace estimation method in uniform circular array still has the problem of sharp degradation of performance in the case of low signal-to-noise ratio (SNR). In this paper, another improved method is proposed. The effect of noise on the performance of subspace estimation is greatly weakened for the essential cause of performance degradation. However, this method needs to estimate the noise power. MSWF-class subspace estimation method and uniform circular array root-finding MUSIC algorithm have many combinations. Finally, the performance of multiple combination methods is analyzed by simulation experiments.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.7
【参考文献】
相关期刊论文 前5条
1 于红旗;黄知涛;周一宇;徐欣;;一种不需要特征值分解的MUSIC方法[J];国防科技大学学报;2007年04期
2 黄磊,吴顺君,张林让,冯大政;快速子空间分解方法及其维数的快速估计[J];电子学报;2005年06期
3 黄浩学,吴嗣亮;基于均匀圆阵的信号源DOA和多普勒频率估计算法[J];电子学报;2001年05期
4 吴仁彪;一种通用的高分辨率波达方向估计预处理新方法[J];电子科学学刊;1993年03期
5 殷勤业,邹理和,Robert W.Newcomb;一种高分辨率二维信号参量估计方法——波达方向矩阵法[J];通信学报;1991年04期
相关博士学位论文 前1条
1 赵大勇;阵列信号处理中的DOA估计关键技术研究[D];哈尔滨工程大学;2011年
，本文编号：2150447
本文链接：https://www.wllwen.com/kejilunwen/wltx/2150447.html
