星座图加权匹配的调制方式识别  

Modulation Classification Using Weighted Constellation Matching

作　　者：张路平[1] 王建新[1] 

机构地区：[1]南京理工大学电子工程与光电技术学院,南京210094

出　　处：《应用科学学报》2009年第1期19-23,共5页Journal of Applied Sciences

摘　　要：通过分析现有基于星座图的调制识别法的局限性,提出一种改进方法。首先计算信号的高阶矩,把11种数字信号归为3类,然后作模糊C均值聚类,将得到的重构星座图与相应的理想星座图模型进行比较,通过引入新的代价函数实现多种数字调制方式的识别。实验表明该方法在识别率和识别范围上均有提高,在信噪比为10 dB时平均识别率提高25%以上。Limitation of existing modulation classification based on constellation is studied, and an improved technique proposed. Eleven kinds of signals are divided into three groups with two higher-order moments. Fuzzy C-means clustering is applied to recover the unknown constellations. The recovered constellations are then matched with standard constellations and classified using an improved cost function. Simulation results show that the correct identification rate is high even if there are more constellation candidates, with the average recognition rate being improved by more than 25% at SNR of 10 dB.

关 键 词：高阶矩 星座图 聚类 调制识别 

分 类 号：TN911[电子电信—通信与信息系统]