Signal Denoising And Modulation Pattern Recognition Based On VMD

At present,the automatic identification technology of signal modulation mode has become a hot topic for scholars.The communication environment presents an increasingly complex and intelligent trend,and the signal is easily impressed by a large number of complex noises.Noise ratio is one of the criteria for measuring the quality of a signal.The higher the signal-to-noise ratio,the better the quality of the entire communication system,so it is important to improve the signal-to-noise ratio.In non-cooperative communication,the two parties cannot predict the modulation mode of the transmitted signal,and for subsequent demodulation,automatic classification of the modulation mode is very important.Based on the predecessors,this paper analyzes the generation principles of three quaternary modulation signals 4ASK,4FSK and 4PSK and obtains the theoretical simulation diagram of the time-frequency domain waveform.The influence of Gaussian white noise in the communication system is introduced.The signal are simulated in the environment with signalto-noise ratio of 7dB and signal-to-noise ratio of 20 dB.The paper elaborates a new signal preprocessing method — variational mode decomposition VMD.The VMD algorithm can decompose the original input signal at different frequencies according to a specific number of layers.In this paper,the VMD decomposition of the three types of modulated signals with signal-to-noise ratios of 7dB and 20 dBis respectively simulated.The waveforms and spectrograms of each signal after decomposition are obtained,and the signals can be deeply studied from different levels.The main experimental content of this thesis is to denoise the noisy modulated signal by VMD decomposition combined with energy entropy.Considering the noisy signal with a signalto-noise ratio of 7dB,the signal of different layers is selected by the energy entropy of the decomposed signal components,the noise component is directly eliminated.The second denoising is performed by the amplitude threshold,so that a relatively pure denoising noise can be obtained..The denoising effect of the algorithm is judged from the comparison of the experimental output signal with the time domain waveform of the original signal,the signal-tonoise ratio comparison and the root mean square error value.The feasibility applicability of the algorithm and the three signals of 4ASK,4FSK and 4PSK are verified.At the end of the paper,the paper uses these three kinds of modulated signal components decomposed and denoised by the VMD algorithm to extract the feature quantities that can effectively classify the inter-class patterns.The features are the instantaneous amplitude correlation coefficient and the instantaneous frequency correlation coefficient energy.Through multiple simulation experiments,a small sample of 1000 data is generated,and two series of support vector machines are combined to automatically classify and identify the three types of signals,and the classification recognition rate of the modulated signal after denoising is obtained.Finally,comparing the three signals in different low SNR environments,the VMD and EMD denoising algorithms improve the recognition accuracy and verify the advantages of the algorithm.