运用高阶累积量和SVM的调制自动识别

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。

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

Applying high order cumulants and support vector machines(SVM), an algorith m is proposed for automatic recognition of digital communication signals. Using the forth,sixth and eighth orders of signals, the new algorithm forms five new classification features at first, then identifies eight types of signals by bina ry tree based SVM as the classifier. Computer simulation indicates that the alg orithm outperforms the one based on BT- SVM, and the rate of accurate classifica tion is over 90% when SNR is above 5dB.