Radio signal modulation identification network based on hybrid neural network and implementation method

Abstract

The invention provides a radio signal modulation recognition network based on a hybrid neural network and an implementation method. The method comprises the steps: extracting the feature information of a modulation signal through a convolution layer, carrying out the dimension division of the feature information extracted by the convolution layer, and converting scalar features into vector features, therefore, the proposed hybrid neural network model is enabled to fully extract the spatial features of the modulation signal, then a gating cycle unit layer is utilized to extract the feature information related to time, and the proposed hybrid network model fully combines the feature information of the signal in time and space states. According to the invention, the model can extract spatialfeatures of the modulation signal more comprehensively, and the classification performance of the modulation signal is improved; the model can extract the space and time characteristics of the modulation signal more comprehensively, and the classification performance of the modulation signal is improved.

Description

technical field [0001] The invention relates to the technical field of wireless communication, and specifically relates to a modulation recognition network and an implementation method. Background technique [0002] In recent years, with the continuous update of communication technology, in order to meet different customer needs, signals are transmitted in different modulation methods. Automatic modulation recognition can accurately determine the modulation type of a signal when the modulation information is unknown. [0003] There are two main methods for automatic modulation recognition of signals, one of which is based on the likelihood ratio decision theory. According to the known signal probability distribution, the appropriate likelihood function is selected as the classification basis of the modulated signal. According to the statistical characteristics of the signal, according to the principle of minimizing the cost function, the test statistics are obtained throug...

AI Technical Summary

Problems solved by technology

[0009] In order to overcome the deficiencies of the prior art, the present invention provides a radio signal modulation recognition network based on a hybrid neural network and its implementation method to solve the problem that the signal feature information cannot be fully extracted when the neural network is used alone to extract feature information, resulting in the existence of classification accuracy. The problem of deviation

Images