基于深度残差收缩网络的信号调制类型识别

Signal Modulation Recognition Based on Deep Residual Shrinkage Network

针对信号调制类型识别问题,提出一种基于深度残差收缩网络(DRSN)的识别方法。算法将原始的IQ两路时域信号数据直接输入网络,利用DRSN来学习时域序列中的调制特征以识别信号调制类型。与现有算法相比,该算法的网络输入为原始时域序列数据,特征提取和识别均在网络中进行,避免了人工特征提取的不完备性;借助DRSN的软阈值化和注意力机制,可有效抑制噪声干扰,从而提高网络从含噪声环境中提取有用特征的能力。仿真实验验证该算法的有效性和优越性。

Aiming at the problem of signal modulation recognition,a recognition method is proposed based on deep residual shrinkage network(DRSN).The time series of raw I/Q signal data are directly input into the network,and then the DRSN is used to extract the modulation features from the time series to identify different signal modulation types.Compared with existing algorithms,the network input of the proposed algorithm is the time series of raw I/Q signal data,and the feature extraction and recognition are performed in the network,which avoids the incompleteness of artificial feature extraction.By using the soft thresholding and attention mechanism of DRSN,the noise interference can be effectively suppressed,so as to improve network ability to extract useful features in noisy environment.Simulation results verify the effectiveness and superiority of the proposed algorithm over existing algorithms.