基于概率密度估计盲分离的通信信号盲侦察技术  

Communication signal blind reconnaissance technology based on probability density estimation blind sources separation

作　　者：付卫红[1] 杨小牛[2] 刘乃安[1] 曾兴雯[1] 

机构地区：[1]西安电子科技大学ISN国家重点实验室,陕西西安710071 [2]中国电子科技集团公司第三十六研究所,浙江嘉兴314001

出　　处：《华中科技大学学报（自然科学版）》2006年第10期24-27,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金："十五"国防预研项目;通信抗干扰国家重点实验室基金资助项目

摘　　要：为解决复杂多信号环境下的通信侦察难题,提出一种新的盲侦察技术,采用基于密度估计的盲分离算法(DEBSS)分离出原始信号,然后对分离的各个信号进行后续信号处理.DEBSS算法采用核函数估计法估计出信号的概率密度函数及其导数,以此确定信号的评价函数,然后采用自然梯度迭代算法进行迭代.仿真结果表明,采用该方法在无需任何先验知识的情况下(如载频、信号带宽、调制样式),可以很好地分离出原始的发射信号,为后续信号处理(如分析识别、解调等)奠定基础.该算法可以对任意源信号进行分离,而不管它是超高斯还是亚高斯信号.Ii is difficult that communication reconnaissance is carried out under the circumstances with complex multiple signals. A new blind reconnaissance technology is proposed, which adopts algorithm of density estimation blind sources separation （DEBSS） to separate the original signals and implement following signal procrssing to each of the resulting signals. DEBSS estimate the signal＇s probability density function and its derivation by kernel function. According to them, the score function can be determined. EASI （equivariant adaptive separation via independent） iterative expressions were used to separate the original signals. The simulation result shows that original signals can be separated by adopting this method without any priori information （i. e. carrier frequency, signal bandwidth and modulation mode）, which establish a base for following signal processing, such as signal analysis and identification, demodulating and so on. Advantage of the proposed algorithm is that any source can be separated, whether it is super-Gaussian or sub-Gaussian signal.

关 键 词：通信对抗 侦察 盲源分离 概率密度函数 

分 类 号：TN911.7[电子电信—通信与信息系统]