Research On Modulation Signal Recognition Technology Based On Deep Learning

In the field of wireless communication,modulation recognition is an important part of non-cooperative communication.The diversified channel environment brought about by the development of communication technology has made modulation recognition more difficult.With the advent of deep learning,it has achieved good results in image and speech recognition.Deep learning can solve the difficulty of effective feature extraction.Using deep learning to automatically recognize modulated signals has become an important research direction in the communication field.First,the basic principles of modulation mode and the theory of deep learning are introduced,common deep learning neural network models are introduced,and the advantages and disadvantages of different neural network models are analyzed.Then,an improved modulation signal recognition method of a residual neural network is proposed.By increasing the width,the extracted signal feature types are enriched,and the over-fitting phenomenon caused by the deepening of the depth is avoided.Simulation results show that the proposed method can achieve an accuracy of more than 95% when the signal-to-noise ratio is 10 d B,which effectively improves the accuracy of the automatic identification of modulated signals.Finally,a modulation signal recognition network based on transfer learning is proposed.By fixing part of the original neural network layer,the remaining network layers are trained twice,which realizes the rapid update of the transfer learning neural network parameters,thereby greatly reducing The training time of the entire network.The simulation results show that the modulation recognition model based on migration learning can realize the rapid recognition of modulated signals in the new channel environment,and the recognition rate can reach more than 95% when the signal-tonoise ratio is 15 dB.