A method of modulated signal identification

Abstract

The present specification relates to the technical field of wireless communication, and relates to a modulation signal identification method, the method comprising: obtaining training set data, verification set data and test set data, and obtaining a parameter initialized modulation signal identification model; training the modulation signal identification model based on the training set data until a loss function of the modulation signal identification model converges or a training frequency reaches a threshold value; calculating a prediction accuracy rate of the modulation signal identification model in a training process based on the verification set data, and selecting model parameters when the prediction accuracy rate is the highest as a training result after the training is completed; testing a prediction accuracy rate of the trained modulation signal identification model based on the test set data, and continuing the above model training process when the prediction accuracy rate exceeds a threshold value; and using the modulation signal identification model to identify a modulation signal. The method of the present specification embodiment can accurately extract effective features, and has a high modulation category identification accuracy rate for a signal.

Images