基于四阶累积量矩阵特征分解的宽带测向算法

Wide-band direction finding algorithm based on four-order cumulants matrix eigen-decomposition

针对信号处理领域研究中的宽带测向问题,基于四阶累积量矩阵定义的灵活性,给出了一种四阶累积量矩阵定义形式,利用累积量的盲高斯性进行简化。通过聚焦操作,把各个窄带频率处的阵列输出矢量变换到聚焦频率处,然后求其累积量矩阵。对各个累积量矩阵的平均值进行特征值分解,就得到四阶信号子空间和四阶噪声子空间。应用子空间的方法进行一维谱峰搜索,即可得到来波方向的估计值。实验结果证明了该算法的有效性。

Wide-band direction finding is a popular and difficult topic of research in signal processing domain. This paper defines a four-order cumulants matrix based on the flexible definition of four-order cumulants matrix and simplifies it using the insensitivity of cumulants to additive Gaussian noise. By focusing operation, array output vectors at different frequency bins are transformed to focusing frequency and individual cumulants matrix is got. Direction of arrival is estimated using subspace method after the eigen-decomposition of the average cumulants matrix. Extensive simulation results demonstrate that the algorithm has good performance.