Modulation Signal Recognition Method Based on Curriculum Learning

Abstract

The invention discloses a modulation signal recognition method based on course learning, which mainly solves the problem of low recognition rate caused by signal noise in the prior art. The scheme is: obtain the training modulated signal sampling sequence and corresponding marked data, and preprocess the sampling sequence; construct a deep residual network; use the preprocessed sampling sequence as the input of the deep residual network, and the marked data of the sampling sequence As the modulation type corresponding to the largest component in the output vector of the deep residual network, the deep residual network constructed by training the training strategy of course learning is used to obtain the trained network; the grayscale image of the modulated signal to be identified is used as the trained network Input, the modulation type corresponding to the largest component in the network output vector is the identified modulation type. The invention accelerates the speed of training, reduces the influence of too strong signal noise on the recognition rate, improves the modulation recognition performance in a strong noise environment, and can be used for electronic countermeasures and radio management.

Description

technical field [0001] The invention belongs to the technical field of communication, in particular to a modulated signal identification method, which can be used in electronic countermeasures and radio management. Background technique [0002] The purpose of the automatic modulation recognition task is to detect the modulation type of the received signal and recover the signal through demodulation. Currently common digital modulation signals include amplitude shift keying ASK, frequency shift keying FSK, phase shift keying PSK and quadrature amplitude modulation QAM. Modulation recognition has been widely used in electronic warfare, surveillance, and threat analysis. Among them, the likelihood-based detection method and the feature extraction method are two commonly used automatic modulation recognition methods. The detection methods based on the likelihood ratio mainly include the average likelihood ratio test method and the generalized likelihood ratio test. Although t...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the deficiencies in the prior art above, to provide a modulation signal recognition method based on curriculum learning, to guide the training of deep residual network with the training strategy of curriculum learning, to avoid the network from overfitting the data noise, Make the network learn a more robust model faster, improve the recognition performance in a strong noise environment, and avoid the problem of low recognition rate caused by signal noise in the existing network

Images