论文部分内容阅读
准确估计信源数目是很多高分辨算法得以实现的前提条件。传统的信源估计方法大多是在白噪声背景下,以数据协方差矩阵的特征值序列作为研究对象,按照某种标准设定门限来进行判决估计信源数目。而在实际情况中白噪声很难满足,从降一维特征空间的投影矩阵角度出发,获得数据协方差矩阵的局部数据的投影差值序列,从而有效避开了噪声主要能量的影响,对空间色噪声也有一定的抑制效果。仿真结果表明在色噪声背景下,利用该方法可以在多信源情况下准确估计信源数目。
Accurately estimating the number of sources is a prerequisite for many high-resolution algorithms. The traditional source estimation methods are mainly based on the eigenvalue series of data covariance matrix under the background of white noise, and the threshold is set according to some standard to estimate the number of sources. However, white noise can hardly be satisfied in practical situations. From the perspective of the projection matrix in the reduced one-dimensional feature space, the projection difference sequence of the local data of the data covariance matrix is obtained, which effectively avoids the influence of the main noise energy. Colored noise also has some inhibitory effect. The simulation results show that under the background of colored noise, the method can accurately estimate the number of sources under multi-source conditions.