一种基于改进盖氏圆盘法的信源数估计新算法

A New Algorithm of Source Number Estimation Based on Modified Gerschgorin Disk

经典盖氏圆盘法(GDE)存在最大可估计信源数减少、盖氏圆半径区分不明显以及调整因子依赖人为经验等问题,实际使用中存在一定局限性。对此,提出一种基于改进盖氏圆盘的信源数估计新算法,引入新的酉变换矩阵及特征值加权矩阵,结合鲁棒性信源数判定准则,提升了算法的估计性能和适应性。理论分析及仿真结果均验证了算法的有效性。

The classical Gerschgorin disk algorithm （GDE） has some problems, such as the re- duction of the maximum estimation number of sources, not obvious difference between different Gerschgorin disk radius, and the dependence of adjustment factor on human experience, which are limitations in application. A new source number extimation algorithm based on modified Ger- schgorin disk is proposed, by introducting the new unitary transformation matrix and eigenvalue weighted matrix, alone with robustness criterion, the estimation performance and adaptability are improved. Theoretical analysis and simulation results show the validity of the algorithm.