文档介绍：摘要?Ill Y2066795 现代通信环境越来越复杂,尤其在通信侦察领域所接收到的信号中往往夹杂着各类干扰信号。由于源信号和传输信道的先验信息无法得知,传统的信号处理技术显现出明显的不足,盲源分离技术很好地解决了这一难题,为通信信号盲分离带来了全新的思路和方法。本文在探讨了盲源分离理论知识的基础上,详细分析了两种传统意义下的自适应盲源分离算法:FastICA算法和自然梯度算法,对它们的原理进行了推导,并通过仿真得出FastlCA算法较自然梯度算法收敛速度快,运算量小,处理效率高。针对传统意义下的盲分离技术无法很好的解决实际通信侦察中获取敌方信号的问题,本文从欠定盲源分离入手重点讨论了单通道下的盲源分离技术,并在此基础上改进了一种基于最大后验估计的单通道盲源分离算法,该算法通过ICA理论将源信号构造成一系列的基函数,然后在信号分离阶段通过最大后验估计理论将这些基函数作为先验信息,从而估计出源信号。仿真验证了该算法分离单通道数字基带信号的有效性。关键词:盲源分离自然梯度FastICA单通道最大后验估计 Abstract munication environment has e increasingly sophisticated, especially in munication detectioninwhich receivedsignals are often mixed with avariety ofinterfering the priori information ofthe source signals and the transmission channel isunknown,the traditional signal processing methods have shown obvious source separation(BSS)has solved thisproblem well and brought new ideasand methods munication signalsblindly. Based on theoretic introductions ofBSS,this paper detailedly analyzes two kinds oftraditionaladaptive blind source separation algorithms:the FastlCA algorithmand thenaturalgradient means of thesimulation andafter athorough analysis ofthetwo algorithms,it has proved the FastICA algorithm hasthe fasterconvergence, thesmaller amount ofcalculation andthehigherefficiency. Due tothe issuethattraditionalseparation algorithms couldn’t work effectively in intercepting enemy signals inthe munication detection,this paper focuses on discussing thetechnology ofBSS inthesingle-channel based on the underdetermined blind source a single-channel blind source separation algorithm based on maximum aposteriori estimation algorithm could firstly construct the source signals into a series ofbasis functions through theICA theory and then inthephase ofsignalseparation estimate thesource signals through maximum posteriori theory with the basisfunctions as priori resultshaveproved theefficiency oftheimproved algorithm in separating s