基于半监督聚类理论的MQAM信号的调制识别

Modulation Recognition of MQAM Signals Based on Semi-supervised Clustering Theory

在MQAM信号的调制识别中,传统聚类算法的聚类点数不准确,算法的迭代次数多且误差平方和函数曲线不平滑.针对此问题,提出了一种基于半监督聚类理论重构MQAM信号星座图的调制识别方法,通过标记部分样本点来指导隶属度及聚类中心的更新,再结合支持向量机(SVM)分类器,实现不同阶数MQAM信号的识别.仿真结果表明,该算法对MQAM信号的识别率大于90%,迭代次数少,误差平方和函数曲线平滑.

In the modulation classification of MQAM signals, clustering points based on traditional clustering algorithm is not accurate. The number of iterations of the algorithm is more and the error sum of squares func- tion curve is not smooth. To solve this problem, this paper presents a MQAM signal modulation recognition method based on semi-supervised clustering theory to reconstruct signal constellation diagram. By marking some sample points to guide the membership and updates of the cluster centers, combined with SVM classifica- tion, the different levels of MQAM signal＇ s recognition are realized. The simulation results show that the algo- rithm for MQAM signal recognition rate is greater than 90% , has less iteration and the error sum of squares function curve is smooth.