| 时变混合系统的在线FastICA算法 |
| |
| 引用本文： | 陈海平,张 杭,路 威,张 江.时变混合系统的在线FastICA算法[J].通信技术,2014(2):136-140. |
| |
| 作者姓名： | 陈海平 张 杭 路 威 张 江 |
| |
| 作者单位： | [1]解放军理工大学通信工程学院,江苏南京210007 [2]总参第63研究所,江苏南京210007 |
| |
| 基金项目： | 国家自然科学基金：欠定条件下基于扰信分离的信干比增强方法研究（No.61001106） |
| |
| 摘 要： | 现有的多数盲源分离（BSS，Blind Source Separation）算法都是假设混合系统是时不变的，然而在实际的通信系统中混合系统常常是时变的。传统的快速不动点（FastICA）算法具有快速收敛的优点，但是不能直接用于处理混合系统时变的盲源分离问题。为了提高盲源分离算法的收敛速度和对时变混合系统的跟踪性能，改进了传统FastICA算法，将混合信号分段，在各段样本中估计峭度并采用批处理的方法进行分离。仿真实验表明，改进后的FastICA算法能在时变环境中跟踪混合系统的时变，并能有效地抗多音干扰。
|
| 关 键 词： | 盲源分离 时变 快速不动点算法收敛速度在线算法 |
On-Line FastICA Algorithm for Time-Varying Mixing System |
| |
| CHEN Hai-ping,ZHANG Hang,LU Wei,ZHANG Jiang.On-Line FastICA Algorithm for Time-Varying Mixing System[J].Communications Technology,2014(2):136-140. |
| |
| Authors: | CHEN Hai-ping ZHANG Hang LU Wei ZHANG Jiang |
| |
| Affiliation: | 1. Institute of Communication Engineering, PLA University of Science and Technology, Nanjing Jiangsu 210007, China ; 2. No. 63 Research Institute, PLA General Staff Headquarters, Nanjing Jiangsu 210007, China) |
| |
| Abstract: | Most existing BSS（ Blind Source Separation） algorithms are developed by assuming that the mix- ing matrix is invariable. However, the mixing matrix is commonly time-varying in practical communication system. The traditional fast fixed-point（FastICA） algorithm cannot directly separate the sources mixed in time-varying system, although is very fast in convergence. Thus in order to accelerate the convergence rate and improve the tracking performance for time-varying system, the FastICA algorithm is modified by divid- ing samples into several subsections, thus to estimate the kurtosis and separate the sources off-line in each subsection. Simulation results indicate that the modified FastICA algorithm could track the time-varying system effectively. |
| |
| Keywords: | blind source separation time-varying FastICA convergence rate on-line algorithm |
| 本文献已被 CNKI 维普 等数据库收录！ |
