A Research On Detection And Identification Algorithm Of Radio Co-channel Interference

With the development of large-scale network monitoring activities,individual identification of wireless communication device signal sources has become an important research direction in wireless communication security.In this thesis,for the recognition of wireless communication devices in the process of single feature recognition,the recognition rate is low and the ideal state cannot be achieved.Two schemes for multifeature extraction and fusion are designed.Among them,the characteristics of RF signals from different signal sources are transformed through the statistical characteristics and high Extraction and fusion of order spectrum features to improve the accuracy of classification and recognition.The main work of this article is as follows:1.An optimized combination algorithm based on Local Mean Decomposition(LMD)--FM-ELMD algorithm is proposed to extract the RF fingerprint features of the signal.First,the local mean decomposition algorithm is implemented and simulated,and the analysis is based on the simulation data decomposition.The modal aliasing phenomenon and the phenomenon that the local mean decomposition algorithm is used in the actual signal decomposition process are analyzed.s reason.On this basis,a local mean decomposition combined optimization algorithm,FM-ELMD algorithm,is proposed to alleviate the modal aliasing problem caused by dense modal and discontinuous events,and to a certain extent reduce the modal aliasing problem.The effect on the signal decomposition,thus improving the effectiveness of the actual signal decomposition results.Then the box-dimension feature extraction is performed on the decomposed function function(Product Function,PF),which can reduce the complexity of data calculation.2.Two schemes for extracting RF fingerprints based on bispectrum are proposed: the first is to extract bispectrum image features,calculate the bispectrum of the collected signal and save it as an image,and extract the bispectrum Gabor features,which The operation is converted to the operation of the image in the graphics domain;the second one takes the high-order spectrum of the extracted signal as a feature and calculates the rectangular integral bispectrum of the signal.The spectral route proposes an optimization plan.Finally,based on the extraction of bispectral feature dimension is too large,is not conducive to the calculation,select the main core component analysis(KPCA)algorithm for dimensionality reduction.3.Two schemes for fusion of steady-state transform domain signal features and higher-order spectral signal features based on the Deep Canonical Correlation Analysis(DCCA)algorithm are given.Combined with radio signal experiments,on the one hand,the local mean decomposition of box dimension and double spectrum Gabor operator feature fusion,on the other hand,local mean decomposition of box dimension and rectangular integral bispectrum(SIB)feature fusion.The results show that after the fusion of box-dimensional features and Gabor features of the local mean decomposition of the walkie-talkie,the recognition rate is improved by 4%-10% than before fusion respectively,the average recognition rate reaches 98.1%,and the average false judgment rate is 0.34%.After the fusion of box dimension features and SIB features of the local mean decomposition of the interphone,the recognition rate was improved by 11.55%-18.55% compared with before fusion respectively,the average recognition rate reached 99.65%,and the average misjudgment rate was 0.1%.