Research Based On Individual Recognition Of Communication Broadcasting Station

The technology of individual identification of communication station can be utilized to detect and obtain the "Fingerprint" feature of the radiation source.In the very beginning of the thesis,the mechanism about the generation of "Fingerprint" feature has been analyzed via the communication station's hardware feature.Based on the domestic and international development trends of feature extraction,feature selection and feature classification,the research content and the structure of the thesis have been discussed.And the theories of fourier analysis,wavelet analysis and Shearlet analysis have been studied as the basis of the individual identification.Based on the ambiguity function analysis method,wavelet analysis and Shearlet analysis have been used for the performance improvement of individual identification of communication broadcasting station,which are excellent in the time-frequency analysis.In order to obtain better feature,the piecewise cubic B-spline wavelet,the db wavelet and Shearlet analysis have been used for the instantaneous autocorrelation function of signal.The ambiguity function analysis method is different from the improved methods,because the fourier analysis is used.And the simulations have been conducted under MSK modulation and PSK modulation.The simulation results showed that the effect of the modulation method for improved method is small,while the effect of ambiguity function method are greatly influenced by the modulation method.In the MSK modulation,the results show that the B-spline wavelet analysis method is better and more stable than the traditional ambiguity function method.Besides,the db4 wavelet analysis method,the B-spline wavelet analysis method and the Shearlet analysis method have better anti-noise performance than the ambiguity function method.Finally,the dimensionality reducing and the classification of signal characteristics have been analyzed.In order to avoid the "dimension of disaster",in the signal classification,reducing the dimension is needed,and then a good classifier for classification needs to be concerned,which will play an important role in upgrading the classification effect.After analyzing the Fisher reduction method,KNN classifier and SVM classifier,the influence degree of different classifiers on the final recognition rate have been shown via experimental simulation.