Deep Learning-based Research On Modulation Pattern Recognition Method Of Communication Signal

Faced with a complex electromagnetic environment,the modulation modes of communication signals are becoming increasingly complex,resulting in that existing methods of identifying modulation modes of communication signals cannot accurately and quickly identify the modulation modes of communication signals.The existing recognition methods mainly include the extraction of feature parameters and the selection of a classifier.The recognition rate depends on the feature parameter extraction of the communication signal,and the steps are complicated.In the case of a low signal-to-noise ratio,the noise has a great influence on the accuracy of the characteristic parameters,so that the existing recognition methods cannot well realize the recognition of the modulation mode of the communication signal.Aiming at the problem of fast and accurate identification of communication signal modulation modes,this paper uses a convolutional neural network structure as a modulation identifier model to implement a modulation algorithm based on deep learning.First,research and analyze the generation mechanism of communication signal modulation types and deep learning related theories.The network structure of convolutional neural networks in deep learning,commonly used activation functions,and optimization algorithms are introduced in detail.Signal modulation pattern recognition method provides a theoretical basis.Then,starting from the time-frequency domain of the signal,the algorithm for generating the time-frequency characteristic image of the communication modulation signal is studied.Using different time-frequency analysis methods,through comparison and analysis,a time-frequency analysis method that can highlight the differences between different signal modulation modes is selected,and the time-frequency map is converted into a digital image that can be processed by deep networks.The modulation signal image generation algorithm based on time-frequency features is used to generate training set and test set samples required for deep learning network training.Finally,the optimal model parameters are obtained through the design of the modulation recognizer based on the convolutional neural network and the performance analysis of the convolutional neural network model.At the end of this paper,the optimal model parameters are obtained based on experimental experience,and modulation mode recognition and performance analysis are performed on the simulated communication signals and the measured communication signals,respectively.Experiments show that the convolutional neural network-based communication signal modulation pattern recognizer has good recognition performance for the five types of modulation signals studied,and has important reference value for the development of communication signal modulation pattern recognition field methods.