论文部分内容阅读
一般的盲信号处理方法常忽略噪声的影响,而实际问题中噪声的影响是存在的。本文主要讨论了在协方差矩阵未知的加性高斯噪声中混合系数的盲估计问题。以最大似然估计为基础,本文提出一种求解参数的最优化算法,并给出了混合矩阵和协方差矩阵的计算式。采用高斯混合模型(GMM)来逼近源信号的概率密度函数,简化了算法中的积分,导出了一种实用的期望最大算法(EM)算法迭代式。计算机仿真结果表明,算法不仅能稳定收敛,而且在低信噪比下的性能也很好。
The general blind signal processing method often neglects the influence of noise, but the actual problem has the influence of the noise. This paper mainly discusses the problem of blind estimation of mixing coefficients in additive Gaussian noise with unknown covariance matrix. Based on the maximum likelihood estimation, this paper presents an optimization algorithm to solve the parameter, and gives the formula of mixed matrix and covariance matrix. Gaussian Mixture Model (GMM) is used to approximate the probability density function of the source signal, which simplifies the integral in the algorithm and leads to a practical iterative algorithm of expectation maximum algorithm (EM). Computer simulation results show that the algorithm can not only stabilize convergence, but also has good performance at low SNR.