利用概率估计和支持矢量机的信号调制分类

Digital modulation classification using probability density estimation and support vector machine

提出了一种常用数字通信信号调制分类算法。针对MASK、MFSK和MPSK调制,选取截获接收机输出信号的瞬时幅度、时频脊线和差分基带信号作为分类特征,利用概率密度估计算法求取分类特征的分布函数,通过构造支持矢量机分类器确定分布函数的峰值个数,从而在多种噪声背景下实现了信号调制类型的自动分类。仿真实验表明,当信噪比大于5 dB时识别率可达80%以上。

A new algorithm is proposed for digital modulation classification. This method analyzes instantaneous amplitude, the ridge of time-frequency representation and difference signals of MASK, MFSK and MPSK signals. Then, probability density estimation algorithm is used to estimate distribution functions of these features. Modulation classification is realized by using support vector machine （SVM） to determine the optimal number of distribution function＇s peaks. Simulation results indicate the discrimination is above 80% when SNR is over 5 dB.