基于UCA的信源个数估计性能改善及其FPGA实现

发布时间：2019-02-09 17:30

【摘要】：在阵列信号处理中,信号源个数的估计在空间谱估计起着关键作用,当估计的信源个数与真实的信源个数不一致时,信号来波方向(DOA)估计算法的性能急剧恶化。因此,准确的估计出信源个数意义重大。当实际噪声不满足白噪声模型时,传统的高分辨方法的性能将会下降。针对现有算法在空间色噪声环境下,信源数接近阵元数甚至超过阵元数时估计性能下降问题,提出了适用于均匀圆阵条件下基于虚拟阵列的信号源个数估计算法,论文的主要内容包括:1.以MUSIC空间谱算法为例,分别研究了信源的错误估计对DOA性能的影响;对常规信源数估计算法简要说明,并验证阵元个数,色噪声相关度、信噪比、快拍数等条件变化对信源个数估计性能的影响。2.针对信源数接近阵元数,现有很多算法因自由度不够导致其估计性能下降,对原始阵列虚拟增加阵元数,提高阵元自由度增加可检测的信号源数目的可能。本文理论分析了内插变换和四阶累积量的虚拟天线技术,通过仿真实验,验证了虚拟阵列增加阵列自由度性能。为了解决四阶累积量中存在冗余信息问题,本文通过阵元坐标矢量信息,直接构造适用于均匀圆阵下去冗余的四阶累积量矩阵,并在白噪声下结合盖尔圆盘准则验证了阵列扩展对信源个数估计性能的影响。3.针对色噪声环境的不确定性,以及有限快拍下四阶累积量矩阵构造缺陷,在信源数接近阵元数时,信号盖尔圆盘和噪声盖尔圆盘没有明显区分,造成调整因子敏感,影响估计性能。利用盖尔圆盘准则和四阶累积量都能一定程度的克服色噪声影响,提出了一种适用于均匀圆阵条件下基于去冗余的四阶累积量矩阵的盖尔圆盘改进算法,避免了人为因素的影响设计了自动调整因子的方法。最后通过理论数据仿真以及实测数据验证了本文算法的有效性,并对本文提出的算法进行了一个复杂度的评估。4.利用Kintex-7系列的FPGA实现信源个数估计算法。分别对每个算法模块进行了FPGA实现分析,最后进行了系统资源评估,并在外场环境实验下进行了FPGA测试分析,验证了系统的正确性。本文提出的方法具体表现在两方面,一方面,在均匀圆阵条件下去冗余后的四阶累积量产生的虚拟阵元能够弥补盖尔圆盘中当信源数接近阵元数时系统自由度不够的缺陷;另一方面,矩阵中噪声特征值发散的问题则可通过对盖尔圆法中半径压缩的改进得到克服。
[Abstract]:In array signal processing, the estimation of the number of signal sources plays a key role in the spatial spectrum estimation. When the estimated number of sources is not consistent with the number of real sources, the performance of the (DOA) estimation algorithm for the direction of signal arrival is sharply deteriorated. Therefore, accurate estimation of the number of sources is of great significance. When the actual noise does not satisfy the white noise model, the performance of the traditional high resolution method will decline. Aiming at the problem of the performance degradation of the existing algorithms when the number of sources is close to or even exceeds the number of array elements in the environment of spatial color noise, an algorithm for estimating the number of signal sources based on virtual array under uniform circular array is proposed. The main contents of this paper are as follows: 1. Taking MUSIC spatial spectrum algorithm as an example, the effect of source error estimation on DOA performance is studied. This paper gives a brief description of the conventional source number estimation algorithm, and verifies the effect of the number of array elements, the correlation degree of color noise, the signal-to-noise ratio and the number of rapids on the performance of the estimation of the number of sources. 2. Since the number of sources is close to the number of array elements, many existing algorithms have the possibility of increasing the number of array elements and the number of detectable signal sources by increasing the number of array elements virtual for the original array due to the lack of degree of freedom. In this paper, the interpolation transformation and fourth-order cumulant virtual antenna technology are theoretically analyzed. The performance of the virtual array to increase the degree of freedom of the array is verified by simulation experiments. In order to solve the problem of redundant information in fourth-order cumulant, a fourth-order cumulant matrix suitable for redundancy of uniform circular array is constructed directly by means of coordinate vector information of array elements. The effect of array expansion on the performance of source number estimation is verified by combining the Gale disk criterion with white noise. 3. 3. In view of the uncertainty of the color noise environment and the defect of constructing the fourth-order cumulant matrix under the finite fast shooting, when the source number is close to the array element number, there is no obvious difference between the signal Gaelic disk and the noise Gaelic disk, which results in the sensitivity of the adjustment factor. Influence estimation performance. Using the Gale disk criterion and the fourth order cumulant to overcome the influence of color noise to a certain extent, an improved Gale disk algorithm based on the deredundant fourth order cumulant matrix is proposed for uniform circular array. The method of automatic adjustment factor is designed to avoid the influence of human factors. Finally, the effectiveness of the proposed algorithm is verified by theoretical data simulation and measured data, and the complexity of the proposed algorithm is evaluated. 4. The number of sources is estimated by FPGA of Kintex-7 series. The implementation of each algorithm module is analyzed by FPGA, and the system resource is evaluated. The correctness of the system is verified by FPGA test and analysis in the field environment. The method presented in this paper is shown in two aspects. On the one hand, the virtual array elements generated by the fourth order cumulant after removing redundancy under the condition of uniform circular array can make up for the deficiency of the system degree of freedom when the number of information sources is close to the number of array elements in Gale disk; On the other hand, the divergence of noise eigenvalues in matrix can be overcome by the improvement of radius compression in Gaelic circle method.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN911.23

【参考文献】