运用高阶累积量和SVM的调制自动识别

Automatic Recognition of Digital Modulation Signals by Applying High Order Cumulants and Support Vector Machines

针对数字信号调制模式识别问题,提出了运用高阶累积量和二叉树支持向量机(SVM)进行自动识别的算法。该算法首先使用信号的四阶、六阶、八阶累积量构造了5个新的分类特征,然后利用二叉树支持向量机分类器实现了8种信号的有效分类。仿真结果表明,该算法优于直接多类分类支持向量机算法,在信噪比大于5 dB时,识别率达到90%以上。

Applying high order cumulants and support vector machines（SVM）, an algorithm is proposed for automarie recognition of dig/tal communication signals. Using the forth, sixth and eighth orders of signa/s, the new algorithm forms five new classification features at first, then identifies eight types of signals by binary tree-based SVM as the classifier. Computer simulation indicates that the algorithm outperforms the one based on BT- SVM, and the rate of accurate classification is over 90% when SNR is above 5dB.