基于实归一化压缩循环统计量的调制识别方法.pdfVIP

Vo l. 17 No. 1 第 17 卷第 1 期 信息工程大学学报 2016 年 2 月 Journal of Information Engineering University Feb. 2016 DOI: 10. 3969/j. issn. 1671-0673.2016.0 1. 009 基于实归一化压缩循环统计量的调制识别方法 孔豫京，黄杂，马金全 (信息工程大学，河南郑州 450ω1 ) 摘要:为解决传统二阶统计量在 Alpha 稳定分布噪声背景下失效的问题，提出了实归一化压缩 函数和实归一化压缩二阶循环统计量的概念并推导其性质。在研究备择信号该类统计量的基 础上，选取实归一化压缩循环谱特定频率和循环频率点处的幅度作为特征参量，利用最小误差 判决分类方法实现信号调制方式的识别。仿真结果表明，该方法在 Alpha 稳定分布噪声和高 斯噪声背景下均有良好的性能，相较于其它基于非线性变换的方法，性能相近且复杂度更低， 鲁棒性更强，灵活性更好。 关键词:调制识别;Alpha 稳定分布;循环平稳;归一化压缩 中固分类号:TN91 1. 7 文献标识码:A 文章编号: 1671-0673 (2016 )01 -0041 -07 Modulation Recognition Methods Based on Real Normalized Compression Cyclic Statistics KONG Yujing , HUANG Yan , MA Jinquan (Information Engineering University , Zhengzhou 45∞01 ， China) Abstract: Traditional second-order statistics become invalid in the Alpha stable distribution noise environment. To solve 由is problem , the concept of real normalized compression function and real normalized compression second-order cyclic statistics is proposed , and its properties 缸e also de- duced. ßased on studying of this cyclic statistics of the altemative signals , the amplitudes of signals real normalized compression second-order cyclic spectrum in specific frequencies and cyclic frequen- cies are extracted as the feature parameters , and the minimum eπor criterion is used as a classifica- tion algorithm to achieve modulation recognition. Simulation results show 也at 由is method has good performance in both Alpha stable distribut