Physical Layer Authentication Technology Based On RF Fingerprint And Channel Fingerprint

With the rapid development of mobile communication technology,Internet of Things technology,and information processing technology,the role of wireless communication in military and civilian data transmission continues to rise.However,the openness of wireless communication makes it more vulnerable to malicious attacks than traditional wired networks,and its security issues cannot be ignored.Traditional access authentication system in the main use of media access control(Media Access Control,MAC)address to control the user's network access rights.However,because MAC addresses can be easily spoofed,its security is a huge potential problem.In recent years,the physical layer authentication technology has been developed,and RF fingerprint and channel fingerprint extracted from the received signal as device characteristics for authentication has the advantage of being difficult to counterfeit,and has been widely studied.Among them,RF fingerprint is the inherent characteristics of wireless devices,with uniqueness,but its identification performance is often vulnerable to the impact of the transmission environment.The channel fingerprint can reflect the spatial location of the device over time,and by incorporating the channel fingerprint into the RF fingerprint authentication method,the recognition rate and robustness of the physical layer authentication system can be further improved.With this as the main research objective,this paper is devoted to the research of authentication methods based on RF fingerprint,channel fingerprint and fused RF and channel fingerprints,which improves the access security of wireless communication through the machine learning method of Gaussian kernel function of support vector machine and the intelligent adaptive authentication method of real-time update,with important theoretical significance and use value.1.A data-driven physical layer authentication method with multi-feature fusion of RF fingerprints is proposed for traditional access authentication systems that are vulnerable to spoofing attacks.The method uses machine learning techniques to fuse features such as modulation offset,frequency offset,constellation trace figure,and differential constellation trace figure extracted from the received signal,and conducts training and authentication to achieve the purpose of improving the authentication performance in scenarios where the device location is more fixed.The experimental results show that the method can effectively improve the recognition rate of the authentication system to about 95%.2.To address the problem that the performance of RF fingerprint authentication method is easily affected by time-varying environment,the variation law of received signal strength(RSS)in channel fingerprint with different locations is studied,and a physical layer authentication method based on RSS vector is proposed.The method uses multiple receivers deployed around the receiver,combines the received signal strength received by each receiver into an RSS vector,reflects the continuity of channel change between the transmitter and receiver with the continuity of RSS vector change,and authenticates by the mutation point detection rule to achieve the purpose of improving the authentication performance under time-varying environment.The experimental results show that the recognition rate of the method is around 90% in the scenario where the sending device is moving.3.An adaptive authentication method with multi-physical layer attribute fusion is proposed for incorporating channel fingerprints into RF fingerprints to improve system authentication performance.The method uses Gaussian kernel functions in support vector machines to reduce the multidimensional attributes of RF fingerprints and channel fingerprints to one dimension to construct an authentication system,and updates the trained model in real time by an environment adaptive method to achieve the purpose of improving the authentication performance of RF fingerprints and channel fingerprints in time-varying scenarios.The experimental results show that the method can effectively improve the authentication performance to more than 96%.