高斯混合模型自适应盲信号分离  

Adaptive blind signal separation based on Gaussian mixture model

作　　者：马诚[1] 李云红[1] 陈锦妮[1] MA Cheng;LI Yunhong;CHEN Jinni(School of Electronics and Information,Xi’an Polytechnic University,Xi’an 710048,China)

机构地区：[1]西安工程大学电子信息学院,陕西西安710048

出　　处：《现代电子技术》2022年第13期35-40,共6页Modern Electronics Technique

基　　金：西安市科技局高校人才服务企业项目(2019217114GXRC007CG008⁃GXYD7.2,2019217114GXRC007CG008⁃GXYD7.8);国家级大学生创新创业训练计划项目(S202010709003);西安市科技计划项目(2019217114GXRC007CG008⁃GXYD7.13)。

摘　　要：信号概率密度函数估计是盲信号分离的关键步骤,其估计的好坏直接影响算法的性能。传统的盲信号分离算法中所用的信号概率密度函数一般只适合轻拖尾信号,而无法准确描述重拖尾信号、冲击脉冲信号的概率特性,使得分离效果较差。针对此问题,提出一种高斯混合模型自适应盲信号分离算法。该算法采用高斯混合模型的概率密度函数估计技术,可以根据高斯核函数理论直接对混合信号的评价函数进行估计,从而实现盲信号分离。文中就轻拖尾与轻拖尾信号的混合,重拖尾与重拖尾信号的混合,以及轻拖尾与重拖尾信号的混合三种情况进行了仿真实验验证,并与Kernel ICA、广义高斯模型、扩展最大熵进行对比。通过不同样本数目的盲信号分离结果可知,文中算法的分离效果较好,具有较高的信噪比。The estimation of the signal probability density function is a key step of blind signal separation,and its estimation effect directly affects the performance of the algorithm.The signal probability density function used in the traditional blind signal separation algorithms is only suitable for light trailing signals generally,and fails to accurately describe the probability characteristics of heavy trailing signals and shock pulse signals,which makes the separation effect poor.To solve this problem,an adaptive blind signal separation algorithm based on Gaussian mixture model is proposed.In the algorithm,the probability density function estimation technology of the Gaussian mixture model is used,by which the evaluation function of the mixed signal can be directly estimated according to the Gaussian kernel function theory,thereby the blind signal separation is achieved.The simulation experiments were carried out to verify the three situations,named the mixture of light trailing and light trailing signals,the mixing of heavy trailing and heavy trailing signals,and the mixing of light trailing and heavy trailing signals.The results are contrasted with those of Kernel ICA,Generalized Gaussian model and extended maximum entropy.According to the separation results of the blind signals with different sample numbers,it can be seen that the separation effect of the proposed algorithm is better,and it has a higher signal⁃to⁃noise ratio(SNR).

关 键 词：盲信号分离 概率密度函数 评价函数 高斯混合模型 轻拖尾信号 重拖尾信号 信噪比 

分 类 号：TN911.7-34[电子电信—通信与信息系统]