| With the development of Internet of things technology,WLAN is more and more widely used.When authenticating the access of WLAN terminal equipment,the traditional authentication method is mainly based on the MAC address of the equipment.This authentication method is easy to be counterfeited and has certain security risks.The RF fingerprint based on the physical layer characteristics is difficult to tamper with,but also has uniqueness and stability,which is very suitable for identification of equipment.This paper selects the WLAN terminal of IEEE 802.11 b protocol currently used in the Internet of things as the research object,proposes three RF fingerprint extraction methods,obtains the advantages of the three RF fingerprints through experimental analysis,and designs a dual device authentication system based on MAC address and RF fingerprint.The main work of this paper is as follows:1.Resampling,frequency offset compensation and other preprocessing operations are performed on the received signal.The frame structure and signal modulation mode of IEEE 802.11 b protocol are analyzed.The sampling frequency is synchronized with the signal transmission rate through resampling,the starting position of each symbol of the signal frame is synchronized through the sliding window method,the starting position of the signal frame is determined through the fixed symbol of the frame preamble,and the overall frequency offset of the signal frame is estimated and compensated by calculating the phase difference of adjacent symbols,The overall phase offset is estimated and compensated by calculating the phase difference between the received signal and the ideal signal.Through the above series of preprocessing operations,the extracted RF fingerprint has stronger stability.2.An RF fingerprint extraction method based on frequency domain signal quotient is proposed.Taking the quotient of two superimposed different signal segments in the frequency domain as the RF fingerprint,the noise influence is reduced by superposition,and the channel influence is eliminated by division.When the signal-to-noise ratio is 30 d B,the classification accuracy of 20 devices is 99.1%;When the signal-to-noise ratio is 5d B,the classification accuracy of 20 devices is92.7%3.A RF fingerprint extraction method based on cross power spectral quotient is proposed.Taking the quotient of two superimposed different cross power spectra as the RF fingerprint,the influence of noise is reduced by the combination of superimposed signal and cross-correlation operation,and the influence of channel is eliminated by division operation.When the signal-to-noise ratio is 30 d B,the classification accuracy of 20 devices is 99.2%;When the signal-to-noise ratio is 5d B,the classification accuracy of 20 devices is 92.5%4.An RF fingerprint extraction method based on LMS adaptive filter coefficient is proposed.The received signal is used as the input of LMS adaptive filter,the local signal is used as the reference input,and the coefficient of the filter is used as the RF fingerprint.The influence of noise is reduced by multi frame data superposition.When the signal-to-noise ratio is 30 d B,the classification accuracy of 20 devices is 100%;When the signal-to-noise ratio is 5d B,the classification accuracy of 20 devices is 93.3%.5.Combined with the different advantages of three kinds of RF fingerprint,a dual device identity authentication system based on MAC address and RF fingerprint is designed.The hardware part of the system consists of a USRP,a computer and several IEEE 802.11 b WLAN routers.The software part includes signal acquisition module,signal preprocessing module,RF fingerprint extraction module and RF fingerprint classification module,which are classified by random subspace algorithm.Combined with RF fingerprint and MAC address,the system can achieve good classification results in the case of low signal-to-noise ratio and large channel changes,which verifies the effectiveness of this scheme. |
