基于支持向量机的数字调制识别算法

　　在非协作通信中，需要对接收的信号进行调制方式的自动识别。在高阶累积量域内构造信号识别的特征向量，采用基于二叉树的支持向量机将识别特征向量映射到高维空间并构造最优分类超平面，实现对数字调制信号的自动识别。该算法不仅结构简单、计算量小，而且解决了样本在低维空间中的不可分问题，具有良好的泛化推广能力。理论分析和仿真结果证明了该算法的正确性和有效性。

　　Abstract:

　　In non-cooperative communication system，it is necessary to recognize the modulation type of the received signals.The algorithm utilizes characteristic vector in high order cumulant domain.SVM based on binary tree maps the characteristic vector into the high dimension space and constructs the optimal separating hyperplane in the space to classify digital modulation signals. The algorithm has simple structure and high computation efficiency， also resolves the non-separable problem in low dimension space and has high generalization performance.The efficiency of the proposed classification algorithm is verified via theoretical analysis and extensive simulations.