Research On Extraction And Recognition Of Radio Frequency Fingerprints Of Wi-Fi Devices In Multipath Environments

Wi-Fi devices have been widely used in daily life.However,in the process of interconnected communication,the traditional method of using the terminal MAC address to identify the device and then implement access control is not secure and vulnerable to identity spoofing attacks.In view of this problem,this thesis proposes that the radio frequency fingerprint of a Wi-Fi terminal can be used as the identity of the terminal to identify and authenticate it.Radio frequency fingerprint is a physical characteristic of wireless communication equipment,which has the characteristics of uniqueness,stability and immutability.However,in a complex communication environment,radio frequency fingerprints are more severely affected by multipath channels,and no feasible solution has yet emerged.The solution to the problem of RF fingerprint extraction and identification under multipath conditions is helpful to expand the applicability of RF fingerprint technology in cyberspace security protection,and has important research significance and practical value.This thesis selects IEEE 802.11 series Wi-Fi devices as the main research object,analyzes the impact of wireless channel multipath effect on RF fingerprints,deduces and experimentally gives the characteristics of RF fingerprints that are less affected by multipath effects,and proposes Corresponding classification and recognition algorithm;at the same time,a set of Wi-Fi access security protection system based on radio frequency fingerprint is designed and implemented to verify the effectiveness of the method proposed in this thesis.The main work of this article is as follows:1.The frame format and relevant details of the IEEE 802.11 series physical layer protocol are studied.Based on this,the influence of multipath channels on the characteristics of radio frequency fingerprints is analyzed,and a series of radio frequency fingerprint characteristics that are less affected by multipath channels are proposed.Including carrier frequency offset,preamble differential curve,preamble differential constellation diagram,and so on.The theoretical derivation of these radio frequency fingerprint characteristics are given,and the effectiveness of these methods are verified by the actual measured signal.2.Aiming at the above-mentioned RF fingerprint characteristics,this thesis proposes methods such as cosine similarity,Pearson coefficient,and K-means clustering to reduce the dimension of feature quantities,and designs a set of ensemble learning classification learning algorithms based on SVM support vector machines.Experimental results show that the algorithm's recognition accuracy for radio frequency fingerprints of mobile devices can reach over 94%.3.An RF fingerprint recognition method based on sequential processing method is proposed.The classification accuracy change mechanism,early warning queue mechanism and variance observation method are used to further improve the recognition accuracy of the algorithm in a low signal-to-noise ratio and in the presence of obstructions.Reduce the false positive and false positive probability of the algorithm.4.Aiming at different signal-to-noise ratio conditions and complex multipath environments,a set of abnormal access detection and attack protection systems for wireless devices based on RF fingerprints was designed and implemented.This system consists of front-end signal acquisition and signal pre-processing module,data processing and classification.The algorithm module and the back-end control admission and alarm module.Through testing,it is concluded that when the signal-to-noise ratio is greater than 15 dB,the recognition accuracy of Wi-Fi devices can reach 85%,and when the signal-to-noise ratio is greater than 20 dB,the system runs better,reaching 90%Recognition accuracy.