Identification Of Communication Radiation Source Based On Complex Degree Of Characteristics

Nowadays,wireless communication technology develops extremely rapidly,and with the rapid development of high-tech and the ever-changing warfare,information warfare will inevitably develop into the mainstream of the future wars,and the communication radiation-source individual identification is one of the key technologies in information confrontation field.Pattern recognition algorithms are mainly used in communication radiation-source individual identification.The general procedures of pattern recognition algorithms are as follows:the signals need to be preprocessed first,including the denoising process or a certain transformation,and then the preprocessed signals are analyzed to extract the related parameters,which can represent the individual characteristics of radiation sources to be saved into the database as the characteristic parameters of the signals.If the intercepted signal characteristics match the signal characteristics in the database,these two signals can be considered from the same radiation sources,so as to achieve the purpose of identification of radiation-source individuals.With the increasingly complex electromagnetic environment of communication,as well as the gradually increased communication signal stypes,how to effectively extract the individual characteristics of the radiation sources at lower SNR becomes a hot issue for scholars all over the world.Aimed at the problem that how to effectively extract the individual characteristics of the radiation sources at low SNR,several new feature extraction algorithms were proposed,and the gray relation classifiers were designed to classify the extracted features.The specific contents are as follows:Since the performance of each feature extraction algorithm needs to be verified by the recognition results of different radiation sources,which means it needs classifiers to classify the extracted features and recognize the signals,this paper introduced a classifier design algorithm first,in order to use the classifiers in the following context.Gray relation theory was used to judge the similar degree of two different discrete sequences by calculating the gray relation values.Comparing with the neural network classifier,it has real-time identification ability,but poor adaptive capacity.To solve this problem,an improved gray relation algorithm was proposed here first and the adaptive ability of the algorithm was improved through adaptively selecting the important degree of each characteristic.The features extracted at low SNR often present characteristics of interval distribution,and to solve this problem,an improved adaptive interval gray relation algorithm was proposed.The simulation results show that,it can realize the purpose of classification of overlapping extracted characteristics at low SNR.Secondly,for the feature extraction module of individual radiation-source identification,the further feature extraction methods based on entropy cloud features and Holder coefficient cloud features were proposed.This algorithm calculated the unstable distribution characteristics of the entropy features and Holder coefficient features at low SNR,which means to extract the three digital characteristics of cloud model,i.e.,mean value,entropy and excess entropy of the characteristic distribution of the first feature extraction results,in order to further extract the distribution features of the discrete signals.Through the secondary feature extraction,it can describe the characteristics of the signals more precisely at low SNR,and then make use of the adaptive interval gray relation classifier to classify the extracted the three-dimensional cloud characteristics,to achieve the individual identification under low SNR environment.Thirdly,individual radiation-source feature extraction method based on improved fractal box dimension was proposed,and the basic theory of one-dimensional fractal box dimension algorithm was improved by getting the derivative of each point of the fitting curves,which constitutes a box dimension feature vector.It can better describe the fractal box dimension characteristics of the signals.Comparing with the traditional one-dimensional box dimension features,it has better recognition results.Finally,individual subtle feature extraction algorithm of radiation sources based on multi-fractal dimension was proposed,and the multi-fractal dimension features of individual radio communication signals or subtle inside noise features were extracted under different noise environments.Thus,it can achieve the purpose of identifying subtle features of discrete signal characteristics under different reconstruction space.