Research And System Implementation Of Wireless Device Authentication Technologys Based On Physical Layer Radio Frequency Fingerprint

With the rapid development of wireless communication technology,the wireless network has been inseparable from people’s daily life.At the same time,the wireless network also faces information security threats that cannot be ignored,especially the problem of user identity tampering and equipment forgery.Traditional security strategies adopt an identification mechanism based on the high-level password protocol.These strategies has certain security risks,such as the existence of key leakage and protocol vulnerabilities.RF fingerprint technology can effectively identify the signal’s device identity by extracting the hardware characteristics of the wireless RF device contained in the wireless signal,so it is widely used in physical identification and access certification of wireless device.However,due to the adverse effects of the multi-diameter effect and noise miscellaneousness in the complicated communication environment,the existing RF fingerprint technology has a certain insufficient robustness and recognition rate.Against this background,this thesis based on the fusion of RF fingerprints,multi-domain fingerprint characteristics,and rapid identification and development of wireless equipment.The main contributions of this thesis are as follows:First,for the problem of traditional RF fingerprints in the multi-diameter environment,it proposes a RF fingerprint recognition method that can resist multi-diameter decline.Wireless device sends radio frequency signals.After identifying the device collection signal,enter the adaptive balancer.Among them,the characteristic graph in the training phase of the radio frequency fingerprint is input to construct a classification model.The equipment entropy and the threshold comparative mechanism are used to compare the legal equipment and illegal equipment.By using USRP and Lab VIEW software radio platforms,a radio frequency fingerprint recognition system based on adaptive equilibrium is established.The experimental test results show that the RF fingerprint recognition of the unbalanced from channels can effectively distinguish legal devices and illegal equipment.The accuracy of identification is higher than 98% and 85%,which ensures the accuracy of radio frequency fingerprint recognition in a multi-diameter environment.Secondly,for the problem of poor classification effects in some single-specific extraction extraction effects,the radio frequency device recognition method based on multi-domain fingerprint characteristics is proposed.Identification equipment collection The feature diagram of the power spectrum,dual spectrum,and constellation trajectory diagram of wireless device launch signals,using the CNN model during the training stage to train and learn to learn and learn from multiple domain characteristics such as power spectrum,dual spectrum,and constellation trajectory.A variety of classifiers are obtained,and the statistical fusion of classification and identification of multiple domain characteristics in the test phase is used to obtain the final recognition results of the test equipment.A multi-domain fingerprint characteristic fusion recognition system based on USRP and Lab View has been established.The experimental test results show that the relative single-special identification method is relatively single-specific.The accuracy rate of identification reaches more than 96%,and the accuracy of illegal equipment has reached more than 92%.Finally,in response to the current problem of high complexity of radio frequency fingerprint classification algorithms,a fast recognition method based on constellation I/Q components is proposed.Wireless device sends radio frequency signals,identifying the device receiving processing wireless signal,of which in the training stage,the K-Means algorithm of the K average cluster is obtained to send the center point of the training equipment to send the signal constellation diagram.The constellation centers of the test device to the correlation coefficient between the constellation cluster center in the training equipment fingerprint database to determine the identity attribute of the test equipment.With the help of USRP and Lab VIEW software radio platforms,a constellation I/Q component system is built.The experimental test results show that the relative CNN algorithm is reported.In the actual scenario of the signal-to-noise ratio than 20 d B,the method can not only ensure the accuracy of identification,but also Time complexity was reduced by 99.75%.This method has obtained a good compromise in both aspects: accuracy and complexity.