基于矢量化差分相位的单分布源解耦二维波达方向估计

在分布源(包括相干分布源和非相干分布源)的二维波达方向估计中,均匀圆阵由于可实现全方位测角、具有较高的分辨率,得到了广泛的应用,然而现有的估计算法均需要谱峰搜索和特征值分解,复杂度较高.针对此问题,考虑单个相干分布源或非相干分布源入射两种情况,提出了一种基于矢量化差分相位的解耦二维波达方向快速估计算法.该算法首先基于空间频率近似模型,证明了任意单个分布源入射时,均匀圆阵中不同阵元接收信号间的差分相位均不受角度扩展参数的影响;基于此特性,通过获取差分相位即可实现中心波达角的解耦合;接下来,提取采样协方差矩阵的严格上三角元素相位,即对应于各阵元间的差分相位,并进行矢量化处理,最终将波达方向估计问题转化为一个最小二乘问题,从而直接得到闭式解,避免了谱峰搜索和特征值分解运算,大幅度降低了复杂度.理论分析和仿真实验表明,所提算法具有较高的估计精度,并且无需角信号分布的先验信息,同时具备较低的计算复杂度和硬件复杂度,有利于复杂环境下阵列测向等工程实践.

Decoupled two-dimensional direction of arrival estimation of single distributed source by vectoring differential phases

In practical applications such as radar, sonar, and mobile communications, transmitted signals are often affected by the scattering and reflection phenomena, which causes the signal energy received by the antenna array to be distributed into a certain space. In this case, a distributed source model will be more applicable. In general, the distributed sources have been classified as coherently distributed (CD) source and incoherently distributed (ID) source, which prove to be suitable for the cases of slowly time-varying and rapidly time-varying channels, respectively.
In this paper, we consider the two-dimensional direction of arrival (DOA) estimation of distributed sources (including CD source or ID source). Specifically, uniform circular array (UCA) is widely used because of its ability to measure full azimuth angle and high resolution. However, the existing estimation algorithms all require spectral peak searching and the eigenvalue decomposition, which can bring a large computational complexity. To solve this problem, a decoupled rapid two-dimensional DOA estimation algorithm is proposed based on vectoring differential phases considering the two cases of single CD source and ID source. Firstly, based on spatial frequency approximation model, it is proved that none of differential phases between the received signals of different sensors in the UCA is affected by angle spread parameters when there is only a single distributed source. Under the premise of such a property, the central DOAs can be decoupled through obtaining the differential phases. Next, we can obtain the phase angles of strictly upper triangular elements in the sample covariance matrix, which correspond to differential phases between different sensors. Finally, by vectoring these differential phases, the central azimuth and elevation DOAs are estimated in the closed form from a least-squared problem, where the spectral peak searching and eigenvalue decomposition can be avoided, hence the computational complexity is reduced greatly. Theoretical analysis and simulation results show that the proposed algorithm has higher estimation accuracy and does not require prior information about the distribution of angular signals. With both low computational complexity and low hardware complexity, the proposed algorithm is beneficial to the engineering practice of array direction finding in complex environment.