基于支撑矢量机的调制制式识别算法  

Identification of Modulation Signals Based on SVM

作　　者：王玉婧[1] 王亚平[2] 赵林靖[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]西安电子科技大学软件学院,陕西西安710071

出　　处：《现代电子技术》2007年第4期192-194,共3页Modern Electronics Technique

摘　　要：调制制式识别是信号自适应传输系统中的一项重要技术。研究表明,传统调制制式识别在训练样本有限情况下性能不佳。而基于支撑矢量机的调制分类器采用结构风险最小化原则,在样本有限情况下仍能达到较好性能。通过分析和研究,首次提出结合高阶累积量和多尺度小波分解两种特征的调制制式识别算法。该算法结合高阶累积量对调相信号以及小波特征量对多载波信号识别的突出优点,实现了多种信号的一步识别。通过对该算法在多种常见信号上的应用进行分析和仿真,证明其性能优于传统调制制式识别算法。Signal modulation identification is one of the important elements most in adaptive transmission. It has been proved that traditional algorithms have the disadvantage in the case of limited train numbers. As Support Vector Machine （SVM） bases on Structural Risk Minimization （SRM）, it can solve the problems. Based on the related research,a new algorithm combining both higher order cumulants and wavelets are proposed firstly. It takes both the advantages of cumulants to phase modulation signals and wavelet to multi - carrier signals. Several normal used modulation type can be classified. And the performance is proved to be good.

关 键 词：调制制式识别 支撑矢量机 高阶累积量 小波 

分 类 号：TN92[电子电信—通信与信息系统]