一种信源数目估计的改进算法

An improved algorithm of source number estimation

在空间色噪声环境下,实际信源数接近于阵元数时,现有源数估计算法结果会产生误差过大,文章提出了一种基于四阶累积量的盖尔圆改进算法。首先对阵列接收的数据构造四阶累积量矩阵,并去除了矩阵的数据冗余,虚拟地增加阵元数,然后利用数据协方差矩阵的特征值梯度大小趋势对盖尔圆半径进行修改,最后完成源数目的个数估计。理论仿真实验表明本文算法较好地解决了实际信源数接近阵列阵元数时,信源数估计误差过大的问题。

In the space color noise environment, when the actual number of sources is more than half of the number and close to the elements in the array, the existing algorithms can produce large errors, and this paper proposes a new algorithm based on virtual array. Firstly, the data received by the antenna array is used to preprocess the data to construct four order cumulative matrix.Then according to the four order cumulant matrix of law,it increases the element numbers of the antenna array by removing the data redundancy of four order cumulant matrix. Finally, gets the source numbers by the Geschgorin Disk Estimator（GDE） and the revised rules. Through theoretical and experimental results, It shows that the algorithm can effectively solve the problem of accurate estimation of the number of sources when the number of actual source is close to the array element.