摘要： 信源数估计是空间谱估计中的重要内容，在估计采用不同的判决准则时，往往需要利用信号协方差矩阵的特征值来进行信源数估计。新算法采用数据矩阵的奇异值分解，通过奇异值建立不同判决准则的判决函数。该算法无需进行协方差矩阵估计，也不需要利用奇异值求解特征值，减少了运算量和估计误差。同时，对数据矩阵进行平滑操作，可以解决信号相干性问题。通过数学推导和计算机仿真，证明了算法的正确性。

关键词: 信源数, 奇异值, 数据矩阵

Abstract: Signal source number estimation is very important to spatial spectrum estiamtion. Usually, eigenvalues of the received signals covariance matrix are used in some different detection criteria. The proposed algorithm method replaces the eigenvalue by singular value, which can be calculated by the data matrix directly. Since the covariance matrix estimation and eigenvalues calculation are avoided, the new algorithm reduces the computation load and estimation error. The spacial smoothing of data matrix resolves the coherent signals. The mathematical calculations and computer simulation show the validity of the new alogrithm.

Key words: signal source number, singular value, data matrix

中图分类号: 

• TP301.6