基于PSO-CF的有限目标信号盲抽取算法
Limited blind target signal extract algorithm based on PSO-CF
石文斌 1魏锋 1王大鸣 1仵国锋 1崔维嘉1
作者信息
- 1. 郑州市解放军信息工程大学信息工程学院通信工程系,河南郑州450002
- 折叠
摘要
以最大化规范四阶累积量绝对值为目标函数,解决了因概率密度函数估计过程中激活函数选取不当而带来算法分离性能下降的问题;引入PSO-CF方法进行优化问题求解,防止算法收敛到局部极值,降低算法的实现复杂度.为高效实现从多源混合信号中抽取有限数目的目标信号,提出了一种基于PSO-CF的有限目标信号盲抽取算法.仿真表明,该算法对超高斯、亚高斯及混合型源信号均可分离,算法普适性强,且收敛速度快,分离性能良好.
Abstract
Considering maximizing the absolute value of normalized fourth-order cumulant as the objective function can avoid decline of algorithm performance caused by unsuitable activation function which is used to estimate probability density function.For solving the optimization problem,particle swarm optimization with constriction factor is chosen to prevent algorithm from converging at local extremum,and then reduce the algorithm complexity of achievement.A limited blind target signal extract algorithm based on PSO-CF is proposed in order to extract limited blind target signals from all source signals.Simulation shows that super-Gaussian,sub-Gaussian or mixed source signals can be successfully separated by the algorithm,which has a good applicability,fast convergence speed and great separation performance.
关键词
规范四阶累积量/PSO-CF/目标信号抽取/相似系数Key words
normalized fourth-order cumulant/PSO-CF/target signal extract/similarity coefficient引用本文复制引用
基金项目
国家科技重大专项基金(2011ZX03003-003-02)
国家科技重大专项基金(2009ZX03003-008-02)
"十一五"国家高技术发展研究计划("863"计划)重点项目资助(2009AA011504)
出版年
2013
NETL
NSTL
万方数据