Simulation Research And Implementation Of Automatic Modulation Recognition Algorithm Of Communication Signal

In modern communication systems, in order to improve the bandwidth efficiency,the baseband signal is modulated to a frequency band signal at the transmission end,and then demodulated at the receiving end; in order to increase the transmission rate,varieties of modulation types are applied in the communication. In this case, beforedemodulating a modulated signal, one must knows its modulation type. So, themodulation recognition technology has become the key to effective communication.Through the study and research of the literature at home and abroad in the field ofcommunication signal modulation recognition, pattern recognition method based oninstantaneous parameters feature extraction and decision tree classifier is used, and themodulation recognition algorithms of this thesis are given. There are three algorithms,one is only for6kinds of analog modulation type, one is only for6kinds of digitalmodulation type, and the other is for12kinds of analog or digital modulation type,which respectively uses3,5and7feature parameters.In the extraction of a signal’s instantaneous frequency information, a windowfunction is used to smooth the phase difference sequence, and the estimation effect issignificantly improved. Since the feature parameter aais difficult to distinguishbetween2ASK and4ASK effectively, the feature parameter a42is used, and thedistinction effect is improved. Since the normalization of the instantaneous frequency,the traditional method is to use the symbol rate, but the value of the symbol rate isalways unknown for the recipient, and it will affect the value of the correspondingfeature parameters. What’s more, there is no concept of symbol rate for the analogmodulation signal. Therefore, the mean of the instantaneous frequency is used for itsnormalization, and the feature parameter f42is used for the classification of thefrequency modulation type, and good recognition results are obtained. In the algorithmfor12kinds of analog or digital modulation, compared that the traditional method uses9characteristic parameters, this thesis uses only7of them.In the integrated development environment Code::Blocks, three algorithms areimplemented with C/C++language. The GUI library wxWidgets, the Discrete Fourier Transform library FFTW, and the plotting library wxMathplot are used for thedevelopment of modulation recognition software’s graphical user interface (GUI) andits functionality. The main interface of the software consists of four drawing windowsand a panel (used to set the parameters and display the result). The software supportsthe modulation recognition of frequency band signal and baseband signal, and itrealizes blind recognition for baseband signal.Finally, the modulation recognition software is tested,12kinds of simulatedmodulation signal data and the actually received FM broadcasting signal data are usedrespectively, the amplitude, phase, frequency and spectral curves (collectively calledthe characteristic curves) of the signal are drawn in the drawing windows, and therecognition result of the modulation type of the signal is displayed on the panel.According to the test results, the software achieves the basic function of modulationrecognition and drawing the characteristic curves.