Research On Radio Frequency Fingerprint Extraction And Identification On LTE-V2X Devices

As the concept of "Internet of Everything" is put forward,the Internet of vehicles is gradually developing and promoting as a typical application in the field of transportation.Due to the complexity of the Internet of vehicles environment and high security requirements,more reliable identity authentication technology is needed.As an important physical feature of communication transmitter,RF fingerprint is unique,stable and difficult to clone.As an important research direction of identity authentication in the field of physical layer security,it provides a new idea for the technical specification of Internet of vehicles security.As the Internet of vehicles system is still in the process of development and promotion,and its signal modulation mode is more complex than other standard signals,there is no effective method to extract and identify RF fingerprint of LTE-V2 X terminal equipment.Therefore,in order to propose a stable and effective RF fingerprint,this paper studies the extraction and identification technology of LTE-V2 X RF fingerprint.This paper takes LTE-V2 X physical layer signal in FDD mode as the main research object,focuses on analyzing frame structure of LTE-V2 X Physical Sidelink Broadcast Channel and Physical Sidelink shared Channel,and proposes LTE-V2 X RF fingerprint extraction and classification methods suitable for these two channels respectively.The effectiveness and stability of the proposed method are verified by experiments.The main work of this paper is as follows:(1)LTE-V2 X physical layer protocols and standards are analyzed.In view of the preprocessing problem of RF fingerprint extraction,a preprocessing algorithm including signal frame capture,time synchronization,frequency offset estimation and compensation,phase offset estimation and compensation is proposed,which lays a foundation for the subsequent extraction of high stability RF fingerprint features of terminal devices.(2)Aiming at the problem of RF fingerprint extraction in the Physical Sidelink Broadcast Channel,in order to avoid the influence of data changes on fingerprint extraction,the RF fingerprint extraction method based on the spectrum characteristics of synchronous symbols and the mutual power spectrum characteristics of main synchronous symbols is proposed respectively by using the relatively fixed synchronization symbols in the frame structure.The former can eliminate the influence of noise on fingerprint by superposition or cross-correlation,while the latter can eliminate the influence of noise on fingerprint by cross-correlation.Through experiments,the classifiers with the best effect are selected for RF fingerprint identification,and RF fingerprint extraction and identification are carried out for single frame data of 12 devices.The accuracy of superposition and cross-correlation methods based on synchronous symbol spectrum is 97.65% and 97.50%respectively,and the accuracy of the method based on main synchronous symbol mutual power spectrum can reach 95.98%.It shows that the proposed RF fingerprint extraction method has good discrimination,stability and anti-noise performance.(3)Aiming at the RF fingerprint extraction problem of Physical Sidelink Shared Channel,a RF fingerprint extraction method based on data symbol is proposed.Under the same bandwidth and initial subcarrier,stable RF fingerprint can be extracted from QPSK encoded data symbol.Through the solution of SC-FMD precoding algorithm,data decoding for constellation diagram,and based on the constellation stipple the brightness figure as RF fingerprint equipment,RF fingerprint using convolution neural network to classify and identify,the four-equipment classification and recognition,the recognition accuracy of 95.8%,verify the feasibility of the algorithm,the stability and good equipment degree of differentiation.(4)The experimental equipment and experimental environment used for signal frame extraction in the LTE-V2 X system are described in detail,and the process of extracting RF fingerprint from the original signal is designed,including the capture of signal frame from the original signal,and the preprocessing of signal frame including time synchronization,channel differentiation,frequency offset estimation and compensation,phase offset estimation and compensation.After the signal frame is preprocessed,the RF fingerprint is extracted and identified by classification.The stability of RF fingerprint extraction method of Physical Sidelink Broadcast Channel signal frame in static state is verified through experimental analysis,and the accuracy of the proposed RF fingerprint extraction method is above 94%.The experiment also verified the long-term stability of the RF fingerprint through two data sets with an interval of two months,and the identification accuracy was more than 92%,which reflected the long-term stability of the RF fingerprint extraction method.