基于并行DBSCAN算法的数字调制信号识别  

Recognition of Digital Modulation Signals Based on the Parallel DBSCAN Algorithm

作　　者：赵知劲[1] 顾嘉伟 

机构地区：[1]杭州电子科技大学通信工程学院,浙江杭州310018

出　　处：《杭州电子科技大学学报（自然科学版）》2016年第2期1-7,共7页Journal of Hangzhou Dianzi University：Natural Sciences

摘　　要：利用四阶、六阶和八阶累积量提取2个特征参数识别2ASK,4ASK,QPSK,8PSK,MFSK和16QAM,有效抑制高斯白噪声的影响.利用改进后的并行DBSCAN算法对特征参数集进行聚类,根据高密度簇中心点与特征参数理论值的距离识别该簇的调制方式,再将未归成簇的数据归于最近的聚类簇中,完成所有数据的识别.仿真结果表明,改进后的并行DBSCAN算法能有效提高聚类效率;信噪比大于4dB时,所提出的二个特征参数能够达到98%以上的信号识别正确率.Two feature parameters extracted from four-order,six-order and eight-order cumulants of signals are used to recognize 2ASK,4ASK,QPSK,8PSK,MFSK,16 QAM,which can suppress white Gaussian noise.The data sets of feature parameters are clustered by the improved parallel DBSCAN.The high-density clusters are recognized by the distances of their centers and the theoretical values of the feature parameters,then the feature parameters outside the clusters are put in the closest cluster to achieve the recognition of all the feature parameters.Simulation results show that the improved parallel DBSCAN is more efficient.When the SNR is higher than 4dB,the recognition rates are higher than 98% by using the two proposed feature parameters.

关 键 词：高阶累积量 调制识别 并行 DBSCAN算法 

分 类 号：TN911.6[电子电信—通信与信息系统]