Journal of Systems Engineering and Electronics ›› 2009, Vol. 31 ›› Issue (7): 1577-1580.

• 电子技术 • 上一篇    下一篇

针对任意非高斯信号的快速固定点盲波束形成方法

赵立权1,2, 杨莘元1, 贾雁飞3, 张朝柱1   

  1. 1. 哈尔滨工程大学信息与通信工程学院, 黑龙江, 哈尔滨, 150001;
    2. 东北电力大学信息工程学院, 吉林, 吉林, 132012;
    3. 北华大学电气信息工程学院, 吉林, 吉林, 132013
  • 收稿日期:2008-03-12 修回日期:2008-12-14 出版日期:2009-07-20 发布日期:2010-01-03
  • 作者简介:赵立权(1982- ),男,博士研究生,主要研究方向为盲源分离,盲波束形成.E-mail:zhao_liquan@163.com

Fast fixed point blind beamforming algorithm for arbitrary non-Gaussian signals

ZHAO Li-quan1,2, YANG Shen-yuan1, JIA Yan-fei3, ZHANG Chao-zhu1   

  1. 1. Coll. of Information and Communication Engineering, Harbin Engineering Univ., Harbin 150001, China;
    2. Coll. of Information Engineering, Northeast Dianli Univ., Jilin 132012, China;
    3. Coll. of Electric and Information Engineering, Beihua Univ., Jilin 132013, China
  • Received:2008-03-12 Revised:2008-12-14 Online:2009-07-20 Published:2010-01-03

摘要: 峭度最大化盲波束形成算法的性能受步长调节参数的选择影响很大,尤其是在信道和信号参数未知的条件下,很难选择合适的步长。针对以上问题,提出了两种新的不需要步长调节参数,而且同样适用于任意非高斯信号的快速固定点的盲波束形成算法。首先通过白化对数据进行预先处理,然后以峭度最大化和波束形成器的权值正交化来构造代价函数,采用复数近似牛顿方法对代价函数优化,得到新的盲波束形成算法。与峭度最大化盲波束形成算法相比,该算法误差小、收敛速度快,不需要任何步长调节参数,更适用于信道和信号未知的环境。仿真实验验证了算法的有效性。

Abstract: The step parameters directly affect the performance of the kurtosis maximization blind beamforming algorithm,especially when the information of channels and sources is unknown,so it is difficult to choose step parameters.Two fast fixed point blind beamforming algorithms without any step parameters are proposed for arbitrary non-Gaussian signals.First of all,the dataset is preprocessed by whitening,then the cost function is constructed by using kurtosis maximization and orthogonal weights.Through optimizing the cost function by complex Newton-Like method,the new blind beamforming algorithms are introduced.Compared with the kurtosis maximization blind beamforming algorithm,this proposed algorithms have less error,fast convergence rate and no need of any step parameters,which are more suitable for a unknown channel and source environment.Simulation proves the efficiency of the algorithms.

中图分类号: