Research And Implementation Of Digital Signal Modulation Recognition Algorithm Based On Graph Features

Modulation recognition is a technical means to identify the signal modulation mode and estimate the modulation related parameters in the absence of relevant prior information.In both military and civilian fields such as radio monitoring and electronic reconnaissance,it has always been crucial.Traditional modulation recognition method mainly uses time-domain waveforms and transform-domain features for recognition.It exists some problems such as the identification highly depends on the experience of the identification personnel,low work efficiency,high labor costs,and poor versatility of the algorithm.Because of the improvement of hardware technology and the artificial intelligence algorithm theory,using the artificial intelligence to develop an algorithm with higher recognition rate,faster speed,and more safety and reliability seems more valuable.This thesis graphically characterizes the digital signal characteristics in the field of modulation recognition,input it into the artificial intelligence algorithm,identify the mode of modulation and estimate the relevant modulation parameters.In the modulation recognition part,aiming at the problem that the low recognition rate,based on the difference of constellation charts of different modulation types,an algorithm model based on graph features is proposed.The model uses constellation plot input,uses cluster analysis to preprocess the signal and extract features,which improves the recognition accuracy of the modulation method.In the modulation parameter estimation part,aiming at the problem that the high parameter estimates value of the digital signal under low signal-to-noise ratio,introducing deep learning related algorithms into parameter estimation process,proposing a modulation parameter estimation algorithm based on convolutional neural network.This algorithm reduces the parameter estimation error value and reduces the error fluctuation range.Finally,this thesis uses the Qt framework combined with Python programming software to develop a software program.Qt is mainly responsible for user operations and results display and other front-end interfaces,and the back-end algorithm is partially implemented by Python.The software not only display the traditional analysis functions,but also can call the background modulation recognition algorithm to display the relevant results of modulation recognition.