点击次数：

论文类型： 会议论文

发表时间： 2019-01-01

收录刊物： CPCI-S

页面范围： 353-362

关键字： Blind modulation classification; Complex correntropy; Asynchronous; Non-Gaussian noise

摘要： Blind modulation classification is an essential and fundamental step before signal detection in intelligent communication systems. However, in complicated electromagnetic environment, identifying asynchronous modulated signals remains a challenging task. In order to improve the performance of asynchronous modulation classification in non-Gaussian noise, this paper proposes a novel BMC method based on complex correntropy and Conv1D (one-dimensional convolution neural network), namely CC-Conv1D. First, complex correntropy is employed to extract discriminating features from asynchronous modulated signals, while non-Gaussian noise can be effectively suppressed by complex correntropy. Furthermore, theoretical analysis is conducted to demonstrate the effectiveness of complex correntropy in feature extraction and non-Gaussian noise suppression. Moreover, Conv1D is adopted to identify different modulation schemes due to its merits of recognizing the shape of extracted features with low computational complexity. Experimental implementation is conducted via USRP N210 and USRP 2901, and the results show that our solution can achieve at least 97.5% accuracy in practical wireless communications.