Research On Radio Frequency Fingerprint Extraction Of Wi-Fi Devices

With the popularization of personal mobile communication devices and industrial IoT facilities,conventional security scheme based on physical address and upper-level security protocols has recently revealed its shortcomings.Using physical characteristics of wireless signals,which natively possesses the specialty of stability and resistance to tamper and counterfeit,as a part of a new-generation network security,has proved its theoretical significance and practical value.This thesis is aimed at Wi-Fi devices.802.11 HR/DSSS PHY and 802.11 OFDM PHY technique is analyzed.A generating method of radiofrequency fingerprints is proposed,for each of the two PHY standards.A recognition and authentication system is established based on it.And the effectiveness of these characteristics is proved by the experiment result.The work of this thesis is concluded as follows.1.The physical frame structure and signal modulation schemes of those two physical-layer protocols are analyzed.A frame detection method is proposed for each physical-layer protocol.2.The influence of carrier frequency offset to the signal frame is studied,and a method of extracting a two-dimensional characteristic vector,based on offset of carrier frequency is proposed,for each of the two physical protocols.The classification experiment using this characteristic carries out accuracy over 90 % among a small number(10~15)of devices of the same product model,and over 97 % for some device models.3.For HR/DSSS frames,a method of extracting characteristic of traces of constellations of spread sequence chips is proposed.The classification experiment carries out accuracy over 99 % among all 45 devices.4.For OFDM frames,a method of extracting characteristic of differential spectrum of preambles is proposed.The classification experiment carries out accuracy over 99.4 % among all 55 devices,and the accuracy is maintained after the channel environment is changed.The authentication experiment carries out EER of around 4 %,which employs Euclidean distance judgement method.5.A wireless classification and authentication system prototype is designed and established.