非平稳噪声背景下DOA分步估计方法研究

DOA Ordinal Estimation Method in Nonstationary Noise Background

针对相干和非相干信源同时存在的情况,结合斜投影理论和互相关矢量Toeplitz矩阵重构(CVTR)的方法,提出一种新的信源DOA分步估计方法。该方法把相干信源和非相干信源分开分辨,对相干信源用CVTR方法来恢复为满秩,避免了常规平滑算法阵列孔径损失大、运算量大的缺点;用斜投影算子而非差分方法对信源进行分离,不受相关矩阵须为Toeplitz结构的限制,使得算法可适用于任意阵列结构形式。这种分步分辨思路可有效增强信源过载能力,同时在互相关矢量Toeplitz矩阵重构过程中,可把非平稳噪声协方差矩阵转换成白噪声结构,使得算法对非平稳噪声有较好的适应能力。

A new ordinal estimation method combining oblique projector and cross correlation vectors Toeplitz reconstruction （CVTR） algorithm is proposed when uncorrelated, partially correlated and coherent sources coexist. With this algorithm, the coherent and incoherent sources are estimated separately The coherent sources are renewed to full rank with CVTR method to avoid the significant loss of array aperture and high computation of the conventional smoo differencing method is used to make the algorithm work thing with method. The oblique projector instead of out the restriction of Toeplitz configuration Therefore, the algorithm is suitable for arbitrary array structure. This method can obviously improve the overload capability ti resolve sources, and along with the matrix reconstruction, the nonstationary noise is transformed to white noise. As a result, the method discussed can fully eliminate nonstationary noise.