基于主成分分析和独立成分分析的调制分类算法

Modulation Classification Algorithm Based on PCA and ICA

基于主成分分析(PCA)和独立成分分析(ICA),提出了一种新的调制分类算法。算法采用PCA对样本数据降维、去除冗余成分,采用FastICA方法提取分类特征;采用支持矢量机(SVM)作为分类器,以解决数据在低维空间中的不可分问题。该算法具有较低的复杂度和较高的训练速度。仿真表明,与最大似然(ML)算法相比,算法仅具有1.8 dB的信噪比损失,在Rayleigh慢衰落信道和中速运动的条件下,算法对5种QAM调制类型具有较好的分类性能。

A principal component analysis （PCA） and independent component analysis （ICA） based modulation classification algorithm is presented. The samples are first processed by PCA to reduce their dimension and eliminate their redundancies, and then the classification features are obtained by the FastlCA algorithm. The Support Vector Machine（SVM） is applied to solve the non-separable problem in low dimension space. The algorithm is less complex computationally and has faster classifier training speed compared with other algorithms. The exten- sive simulation results show that the proposed algorithm has only 1.8 dB SNR loss, and exhibits better classification performance under Rayleigh channel and medium movement condition.