| 信源数动态变化的一种稳健超定盲分离算法 |
| |
| 引用本文： | 李慧,沈越泓,陈士周.信源数动态变化的一种稳健超定盲分离算法[J].军事通信技术,2010(2). |
| |
| 作者姓名： | 李慧 沈越泓 陈士周 |
| |
| 作者单位： | 解放军理工大学通信工程学院研究生2队;解放军理工大学通信工程学院无线通信系;解放军理工大学通信工程学院研究生4队; |
| |
| 摘 要： | 针对大多数现有盲分离算法都假定信源数已知,分析了信源数未知且动态变化时的超定盲分离(接收天线数多于信源数)算法,通过奇异值分解在线估计信源个数,并修正自适应神经网络的维数。文章在经典Cichocki-Unbenauen算法中加入动量项,设置可变步长,使分离性能得到较大的改进,同时避免训练陷入局部最小。与文献1]新提出的ANA算法相比,仿真结果证明本文算法具有更快的收敛速度和更小的稳态误差。
|
| 关 键 词： | 超定盲分离 奇异值分解 自适应神经网络 动量项 |
Robust Over-determined Blind Separation Algorithm with Dynamic Source Number |
| |
| LI Hui,SHEN Yue-hong,CHEN Shi-zhou.Robust Over-determined Blind Separation Algorithm with Dynamic Source Number[J].Journal of Military Communications Technology,2010(2). |
| |
| Authors: | LI Hui SHEN Yue-hong CHEN Shi-zhou |
| |
| Institution: | LI Hui1,SHEN Yue-hong2,CHEN Shi-zhou3(1.Postgraduate Team 2 ICE,PLAUST,Nanjing 210007,China,2.Department of Radio Communication ICE,3.Postgraduate Team 4 ICE,PLAUST) |
| |
| Abstract: | Given that the source number is assumed known in most present blind separation algorithms,the paper dealt with over-determined blind separation with source number unknown and changing dynamically.The on-line estimation of the source number obtained by Singular Va-lue Decomposition(SVD) was followed by the dimension amendment of the adaptive neural network.The paper added momentum term to the classical Cichocki-Unbenauen algorithm,and set changeable step,so as to improve separation performance and avoid the ... |
| |
| Keywords: | over-determined blind separation singular value decomposition adaptive neural network momentum term |
| 本文献已被 CNKI 等数据库收录！ |
|
