Noniterative DOA Estimation Algorithms of Noncircular Signals in Nonuniform Noise Environment
Noniterative DOA Estimation Algorithms of Noncircular Signals in Nonuniform Noise Environment
复制标题
DOI：
10.1109/access.2020.3031213
复制
发表时间：
2020
期刊：
影响因子：
3.9
通讯作者：
Yu Qiao;Han Wang;Xianpeng Wang;Mengxing Huang;Liangtian Wan;Diyang Wu
中科院分区：
文献类型：
--
作者：
Yu Qiao;Han Wang;Xianpeng Wang;Mengxing Huang;Liangtian Wan;Diyang Wu
提出了两种非均匀噪声环境下非圆信号的非迭代波达方向估计算法。与主流的非均匀迭代算法不同，本文提出的算法无需迭代和凸优化处理，即可在非均匀噪声环境下有效地实现DOA估计。在非均匀噪声直接去除（DRONN）方法中，通过对阵列输出协方差矩阵的特殊处理来估计噪声子空间，而且不需要估计噪声协方差矩阵。非均匀噪声分段估计（PEONN）方法首先估计噪声协方差矩阵，然后利用DRONN方法估计噪声子空间，最后利用广义特征分解（GED）估计噪声协方差矩阵。这两种方法都能有效地抑制非均匀噪声的干扰，并且不需要迭代处理就能准确地估计出DOA。此外，这两种方法均采用降维非圆多信号分类（RD-NC-MUSIC）算法进行DOA估计，无需复杂的二维空间搜索，有效降低了计算复杂度。仿真结果证明了这两种方法的有效性。
In this paper, two noniterative direction-of-arrival (DOA) estimation algorithms of noncircular signals in nonuniform noise environment are proposed. Different from the mainstream nonuniform iterative algorithm, the algorithms we proposed in this paper could attain DOA estimation effectively in nonuniform noise environment without iterative and convex optimization processing. In the direct removal of nonuniform noise (DRONN) method, the noise subspace is estimated by using special processing of the array output covariance matrix, moreover, it does not require to estimate the noise covariance matrix. On the other hand, the piecewise estimation of nonuniform noise (PEONN) method first estimates the noise covariance matrix, and the noise subspace used in this process is estimated by using the DRONN method, then the generalized eigendecomposition (GED) is used to estimate the noise covariance matrix. The above two proposed methods are able to suppress the interference of nonuniform noise effectively, and accurately estimate DOA without iterative processing. In addition, the two proposed methods use the reduced-dimensional noncircular multiple signal classification (RD-NC-MUSIC) algorithm to estimate DOA without complex two-dimensional spatial search, and they can effectively reduce the computational complexity. The effectiveness of the two proposed methods are proved via the simulation results.