Research On Extraction And Recognition Of Mobile Terminal RF Fingerprint

With the development of wireless communication technology,the security of wireless networks has attracted more and more attention,especially the identity authentication of access devices is an important part of ensuring the security of communication systems.Traditional wireless based on encryption and decryption algorithms and security protocol algorithms The communication security system is built on the data link layer and above.People begin to pay attention to identity authentication from the lower layer perspective-the physical layer.Therefore,the uniqueness of the physical layer characteristics of the communication equipment and the characteristics that are difficult to forge are wireless.The authentication security of communication is of great significance,and it is worth studying in depth.The research object of this paper is the GMSK signal sent by the GSM mobile device.The modulation method and physical layer structure of the GMSK signal are analyzed.A signal feature generation method for the training sequence is proposed,and it is verified by experiments under multipath and noise conditions.It is used for reliability of device identity authentication.The main work of this article is as follows:In order to extract the steady-state fingerprint of the GMSK signal,the signal modulation principle and burst structure in the GSM standard are studied,the training sequence in the middle of the burst is selected as the target signal,and an algorithm for extracting the training sequence using complex correlation is proposed;An RF fingerprint generation method based on anti-Gaussian filter differential constellation trajectory is proposed,which eliminates the inter-symbol interference introduced by the Gaussian filter and improves the stability of the constructed RF fingerprint.If the signal-to-noise ratio is greater than20 d B,the classification accuracy of the equipment can be more than 90In order to improve the classification and recognition rate of RF fingerprints,this paper attempts to extract the device features of the differential constellation trajectory from the perspective of imaging,and proposes a method of constructing RF fingerprints based on the image contour.The accuracy of classification is about 84%,when the signal-to-noise ratio is greater than 30 d B,the accuracy can reach more than 93%.Although the expected results are not achieved,it enriches the angle and method of RF fingerprint extraction;In order to enhance the stability and robustness of GMSK RF fingerprints,this paper studies the carrier frequency offset and instantaneous phase response statistical characteristics of GMSK signals to construct multidimensional fingerprint features,and obtains 98% classification in the test using SVM classifier Accuracy.