Research On Individual Identification Of OFDM Communication Radiation Source Based On Longformer Network And I/Q Imbalance

Since the 21st century,the development of sensor networks and mobile communications has transformed the information society from the electronic era to the Internet of Things era,which has spawned countless communication radiation source devices,making wireless spectrum resources more tense,and greater threats to wireless network security.Individual identification of communication radiation sources is the process of identifying individuals of different radiation sources by extracting the features in the communication signal that can be used to identify the identity of the communication device that sends the signal,which is of great significance for improving the security of wireless communication systems.As an efficient transmission technology,Orthogonal Frequency Division Multiplexing(OFDM)has the advantages of high spectral efficiency and suitable for broadband transmission,and is widely used in broadcast audio and video fields and civil communication systems.At present,in the actual complex electromagnetic environment,the research results of OFDM communication radiation source individual identification technology are few and the recognition rate is low.In this paper,the deep learning method is introduced into the field of OFDM communication radiation source individual identification,and a method for individual identification of OFDM communication radiation sources based on Longformer network and I/Q imbalance is proposed.The main research contents of this paper are as follows:1.The generation mechanism and processing method of individual fine features of OFDM communication radiation source is introduced.The basic principle of OFDM is analyzed;the nonlinear mechanism of devices in the communication radiation source transmitter is analyzed,and the basic structure model of the communication radiation source transmitter is given;the basic theory of high-order spectrum in modern signal processing is analyzed.2.The basic theory of deep learning and Longformer network is introduced.The main applications and methods of deep learning are expounded,the relationship between deep learning related technologies is given,and the characteristics of the basic neural network model,training process and activation function are analyzed.The relevant theory of the Longformer network is emphatically expounded,and the basic structure,main characteristics and related operations of the Longformer network are introduced in turn.3.An individual identification algorithm of OFDM communication radiation source based on Longformer network and symbol synchronization information is proposed.Firstly,the influence of the I/Q mismatch of the quadrature modulator on the OFDM signal is analyzed,and then the individual characteristics of the radiation source based on the quadrature mismatch are established at the transmitter,and rectangles integral bispectrum are extracted after segmenting the signal according to the symbol synchronization information at the receiver,after position encoding,features are further extracted by the sliding window attention mechanism based on global information in the Longformer network,and finally classified by the Softmax classifier.The recognition effects of different initial learning rates,different numbers of training samples,different signal-to-noise ratio conditions,and different modulation methods are simulated and analyzed.4.The features generated by the mismatch of the I/Q branch filter of the communication radiation source are input into the Longformer network for classification and identification,and its identification performance is analyzed under the conditions of different numbers of train samples and different signal-to-noise ratios.Finally,the identification performance of the transmitter using different types of filters in the multipath channel is simulated and analyzed.The results show that the individual identification method of OFDM communication radiation source based on Longformer network can effectively identify the individual radiation source,and it is found that the feature recognition rate caused by the I/Q mismatch of the quadrature modulator is higher than that caused by the mismatch of the I/Q branch filter.It is suitable as a basis for identifying the subtle features of communication radiation sources.