Application Of Machine Learning Algorithm In Modulation Recognition

Communication signal modulation recognition is an important technology in the field of wireless communication.With the rapid development of wireless communication technology,the modulation types of communication signals are becoming more and more complex,which undoubtedly brings great challenges to the modulation recognition of communication signals.In recent years,with the development of artificial intelligence technology,machine learning algorithms have shown strong generalization ability and classification ability,and have been successfully applied in many fields.This thesis studies the application of machine learning algorithms in the field of communication signal modulation recognition,the main work is completed as follows:(1)The basic principle of digital modulation signals is described,and the common modulation signals are simulated and analyzed;The basic structure and working principle of the support vector machine are introduced,and the structure and characteristics of the convolutional neural network are summarized,which lays the foundation for the subsequent research of modulation signal recognition algorithms.(2)A modulation signal recognition algorithm based on improved VGG neural network is proposed.This algorithm does not need the extracted feature parameters of the signal,but directly takes the sampling points of the signal to be recognized as its input,and uses deep neural network to extract its hidden features for modulation recognition.By deleting the fully connected layer of the original VGG neural network,enlarging the convolutional layer and changing the maximum pooling to the average pooling,the number of parameters and training time of the model are greatly reduced,and a high recognition rate is ensured under low signal-to-noise ratio.(3)A modulation signal recognition algorithm based on random forest model is proposed,which uses the newly constructed high-order cumulants as classification features to realize the effective recognition of more complex modulation signals.Firstly,several high-order cumulants are constructed in time domain,and combined with the instantaneous features and AR spectral analysis,as the feature parameters of signal classification.Then these feature parameters are used as the input of the random forest model for modulation mode classification and recognition.Simulation results show that the proposed algorithm has good recognition performance in the presence of complex high-order modulation signals.