基于特征值分解的频移键控信号检测与识别  

Detection and recognition of FSK signal based on eigenvalues decomposition

作　　者：王甲峰[1] 尹显东[1] 姚军[1] 

机构地区：[1]中国工程物理研究院电子工程研究所,四川绵阳621900

出　　处：《信息与电子工程》2012年第3期312-318,共7页information and electronic engineering

基　　金：中国工程物理研究院科学技术发展基金资助项目(2009B0403043)

摘　　要：提出一种基于协方差矩阵特征值分布特性的频移键控(FSK)信号检测方法以及基于信源数估计技术的FSK识别方法。根据噪声协方差矩阵特征值的直线分布特征,利用直线拟合度构造了信号检测判定准则。给出了常用的信息论(AIC)、最短描述长度(MDL)及盖氏圆盘算法(GDE)信源数判定准则的定义,并根据特征值分布特征提出了一种最大斜率准则(MSR)。仿真表明,MSR准则对于FSK识别更有效。同时,给出了检测及识别参数的设置方法。仿真结果证明了检测及识别方法的有效性。文章将阵列信号处理中的方法应用于单通道信号的分析中,为FSK信号的检测与识别提供了一种新的思路。The eigenvalues distribution of the covariance matrix based detection method and the signal sources estimation based Frequency Shift Keying（FSK） recognition method are proposed. The detection rule is derived using the line comparability based on the line characteristic of the eigenvalues distribution of the covariance matrix of the Gaussian noise. The definitions of Akaike Information Criterion（AIC）, Model Definition Language（MDL） and Gerschgorin Disk Estimation（GDE） rules are introduced. And the Max Slope Rule（MSR） is proposed and the simulations show that the MSR is more effective for FSK recognition, Then the method of setting parameters is given. Simulation results prove that the detection and recognition methods are effective. In this paper, the algorithm of array signal processing is applied to process the single channel signal, which provides a novel method for FSK detection and recognition.

关 键 词：特征值分解 检测 识别 信源数估计 

分 类 号：TN911.23[电子电信—通信与信息系统]