摘要
传统调制识别算法的识别效果受信号的信噪比、成型脉冲形状、采样率等参数影响,适用范围有限。为了提高对具有不同参数的信号的识别正确率,通过对数字调制信号的瞬时特征进行多阶变换,基于变换后的功率变化和分布信息,提出了一种基于瞬时特征谱的联合特征调制识别算法。该算法以集成学习器XGBoost对数据进行训练,能够适应不同条件下的信号数据,使其识别正确率得到明显提高。仿真结果显示了该方法的有效性。
The recognition effect of conventional modulation recognition algorithm is affected by the SNR,shape of forming pulse, sampling rate and other parameters, so its application scope is limited. In order to improve the correct recognition rate of signals with different parameters, through performing a multistage transformation on the instantaneous features of the digitally modulated signal, and based on the transformed power variation and distribution information, this paper proposes a joint feature modulation recognition algorithm based on instantaneous feature spectrum. The algorithm uses an integrated learner XGBoost to train the data and is able to adapt to signal data under different conditions, resulting in a significant improvement in the correct recognition rate. Simulation results indicate the effectiveness of the method.
作者
罗强
刘一兵
赵洋
LUO Qiang;LIU Yibing;ZHAO Yang(Unit 63892 of PLA,Luoyang Henan 471003,China)
出处
《通信技术》
2022年第12期1547-1554,共8页
Communications Technology
关键词
调制识别
瞬时特征谱
联合特征
XGBoost
modulation recognition
instantaneous feature spectrum
joint feature
XGBoost
