基于去特征处理的信源数估计算法

在实际测向系统中,由于阵列中各阵元的相互干扰,阵列接收的噪声为空域相关色噪声,由于系统热噪声和散弹噪声的存在,导致阵列接收的噪声同时包含空域相关色噪声与白噪声。传统针对空域相关色噪声的信源数估计算法在白噪声背景下的估计性能往往下降。针对这一问题,提出了一种基于去特征处理的信源数估计方法。该方法对数据协方差矩阵进行去特征处理,得到一组与特征值相对应的新数据协方差矩阵,利用新数据协方差矩阵在信号和噪声子空间上的投影矩阵构造判决函数,进而实现信源数估计。仿真实验及对比数据表明该方法在包含有白噪声成分的色噪声背景下能够有效进行信源数估计,其突出优势表现在小快拍数、低信噪比条件下仍能准确估计信源数,而没有明显增加算法的运算量。

New source number estimation method based on feature eliminated process

In actual direction-finding systems, the impinging noise is a mixture of white noise and colored noise. The estimation performance of the traditional algorithm used for colored noise tends to decline when it is used under the background of white noise. Accordingly, a new source number estimation method based on the feature eliminated process is proposed in this study. The method performs the feature eliminated process on data covariance firstly. A group of new covariance matrixes according to the eigen-values are obtained. Using the projection matrixes of signal subspace and noise subspace to construct the criterion function,then get the number of signals. Simulation results show that the method proposed can effectively estimate the number of signals under the white noise and colored noise background. The algorithm can estimate the number of sources with high accuracy rate under the condition of small snapshot and low SNR without obvious computation increase.