基于复杂度特征的调制信号识别

调制信号的调制规律能够直接在信号波形上得到反映，本文使用复杂度来定量描述信号的这种有效、简便的波形变化特征，它包括Lempel-Ziv复杂度和分形维数。在Lempel-Ziv复杂度的提取中，本文提出了一种有效的序列重构方法，以此来提高特征的类内集群程度和类间分离程度。在此基础上针对7种调制类型设计了分层结构的组合分类器，从而获得较为满意的识别。计算机仿真实验也证实了本文方法的有效性。

Modulation signals recognition based on the complexity measure

The modulation regularities were indirectly reflected on signal waves. This paper used the complexity measure to quantificationally characterize signals variation signals in order to extract the effective and handy features. It included Lempel-Ziv complexity and fractal dimensionality. When extracted Lempel-Ziv complexity measure, we presented a method of restructuring signals, which could improve collection degree of within and separation degree of between classification. On the basis of it, hierarchical architecture combined classifier was designed for seven modulation types, thereby the good recognition rate was obtained. Simulations verify its good performance.