Research On Automatic Recognition Method Of Wireless Communication Digital Modulation Mode Based On Feature Extraction And Learning

Modulation recognition is an important research field in communication system.It is between signal detection and signal demodulation,which is one of the key technologies to ensure communication safety and reliability.It is widely used in different fields,so it is of great significance to study the automatic recognition of modulation mode.This paper studies how to reduce the influence of noise on characteristic parameters,and designs a better classification method,so as to improve the overall recognition effect.This paper focuses on the following parts:Firstly,To study the a high-order statistical feature extraction technology based on sae-softmax,including the feature extraction method and steps of the technology.This method are combined SAE and SOFTMAX,In combination with the higher order cumulants to realize the fast extraction and identification of modulation mode Characteristic parameters.Secondly,the existing modulation recognition methods are limited in the type of modulation mode and the recognition rate is not high at low SNR.To study the a modulation recognition method based on deep learning.This method takes advantage of the fact that the theoretical value of the high-order cumulant of the zero-mean gaussian white noise is 0.High order cumulants are introduced into the signal analysis process to protect the system from gaussian white noise.At the same time,the introduction of deep learning network structure to complete the characterization of weak features,combined with three characteristic parameters,it can effectively solve the problem of recognition rate decline in low signal-to-noise ratio environment.Thirdly,In view of the fact that prior data in engineering practice is relatively few and the support vector machine has better generalization ability.A modulation recognition algorithm based on SVM feature learning is studied.This method use high-order cumulants to calculate various digital modulation signals and then constructs as the identification parameter.By introducing support vector machine to identify modulation modes.Finally,the seven modulation modes are well recognized.Finally,the experimental results show that the modulation recognition method based on deep learning has a good recognition effect.In gaussian channel environment,the classification accuracy is higher than the existing methods.In different channel environments with low SNR,the model has higher recognition rate.The model is more robust in time,phase and frequency offset.The modulation recognition method based on SVM feature learning has fewer characterization parameters,and can be well applied to the small sample test environment with remarkable recognition effect.