The Modulation Recognition System Based On Feature Extraction

The modulation identification is the way to identify wireless communication signal by certain algorithm with less prior knowledge.With the rapid development of communication technologies and wireless devices,the communication signal modulations are increasingly diversified.Due to signal interference in the environment,it becomes more and more difficult to identify the modulation of the communication signal.Nowadays,the problem that how to improve the recognition rate under low SNR(Signal to Noise Rat io)and other conditions is important key for wireless communication signal modulation recognition.In recent years,modulation is researched and put forward a variety of mature and practical algorithms,including maximum likelihood,parameter statistics and so on.However,these algorithms need to get c parameters in advance or have more requirements on priori parameters,resulting in poor application in practice.So the modulation identification for digital signal is researched in-depth in this paper.The main work and innovations are as follows:Firstly,the digital modulation signals principle of cyclic spectrum is analyzed and studied,and the characteristics of different signals are analyzed.The cyclic spectrum correlation method is used to extract features of six modulation signals,and the modulation recognition decision process is obtained through experiments.Secondly,aiming at the problem that the signal recognition rate is poor with the low SNR ratio,the methods of cyclic spectrum and multilayer perceptron network are proposed to identify signal in this paper.Experimental results show that great recognition effect can be achieved if the cyclic spectrum has a certain degree of discrimination.Thirdly,aiming at the problem of low signal recognition rate without priori parameters,a modulation recognition method based on LSTM is proposed by carrier frequency estimation,signal division,data processing and LSTM identification in this paper.And the recognition rate and recognition speed are obtained to analyze under different condi tions.This method realizes the recognition of the signal when the prior parameters are insufficient,which verifies the availability through experiments.Finally,a modulation identification system is developed to solve the practical problems in modulation signal recognition and facilitate the design of the modulation signal recognition,which can identify signals by cyclic spectrums feature,multilayer perceptron and LSTM.