Modulation Classification Of Communication Signal Based On Deep Learning

Modulation recognition technology is to identify the modulation of the signal in the environment of insufficient prior information.Modulation recognition technology plays a very important role in both military and civil fields.In the military field,in order to obtain the nonpartner signal,modulation recognition technology is needed for efficient and accurate recognition.In the civil field,modulation recognition technology can supervise the electromagnetic spectrum.When the electromagnetic spectrum is illegally occupied by unknown signals,modulation recognition technology can be used to further identify the unknown illegally occupied signals.According to the application of deep learning algorithm,the first is modulation recognition algorithm based on machine learning,the second is modulation recognition algorithm based on deep learning.The original modulation recognition algorithm is difficult to adapt to the complex electromagnetic environment after continuous development,so the deep learning algorithm is introduced into the field of modulation recognition.In this paper,three algorithms based on machine learning and three algorithms based on deep learning are used to simulate and compare the sample data in the same public data set.A convolutional neural network with linear transformation module is proposed to build an end-to-end network model.This method can learn the characteristics of the input sample data more comprehensively.Compared with the classical convolution neural network method,this paper replaces the deep learning two-dimensional real convolution with one-dimensional linear convolution,so that the deep learning architecture can calculate complex convolution operations,and learn the characteristics of the real part and imaginary part of the received signal,so as to improve the accuracy of modulation recognition.Simulation results show that the proposed method can effectively identify the modulation mode of unknown signal,and the accuracy can reach about 80%when the signal-to-noise ratio is greater than 5dB.Compared with the classical convolution neural network,the improved convolution neural network has more obvious improvement.