基于超平面法矢量的欠定盲混合矩阵估计

Underdetermined Blind Mixing Matrix Estimation Based on Normal Vector of Hyperplane

源信号稀疏性差时,基于源信号稀疏特性的欠定盲混合矩阵估计算法,通常先聚类求得混合矢量张成的超平面,然后估计混合矩阵。但此方法涉及运算量较大的超平面聚类,算法效率低。针对这一缺陷,提出了一种新的混合矩阵估计算法。先由所提出的基于梯度法的法矢量更新方法求得超平面法矢量的估计,然后求出混合矩阵。该方法不需要进行超平面聚类,大大降低了运算量,提高了混合矩阵估计效率。仿真结果证明了该方法的正确性和有效性。

When sources are not strictly sparse,the algorithms of underdetermined blind mixing matrix estimation based on the sparsity of sources usually firstly cluster the hyperplanes generated by the mixing vector,and then estimate the mixing matrix.However,this method requires the calculation of hyperplane clustering whose computa-tion load is heavy and efficiency is low.To address this issue,a new algorithm is proposed.First,the normal vector of hyperplane is calculated by the proposed normal vector renew formula based on the gradient method,and then the mixing matrix is estimated.In this way,hyperplane clustering is avoided.The proposed algorithm has lower computational cost and the efficiency of the estimation of mixing matrix is well improved.The simulation results verify the accuracy and the effectiveness of the proposed algorithm.