Identification Of Spoofing Of Satellite Navigation System Based On RF Fingerprint

Global satellite navigation system(GNSS)has already played a significant role for human being with the services of all-weather timing,positioning and navigation in the military and civilian aspects.Unfortunately,the fragility and vulnerability of GNSS is also exposed gradually.Particularly,spoofing is increasingly becoming one of the main threats to the GNSS because of its low power,quiet stealth and easy acquistion.This dissertation mainly studies the anti-spoofing algorithm based on radio frequency fingerprinting of GNSS.The main works of this thesis are as follows:Firstly,the development and the vulnerability of GNSS are introduced.A survey of jamming and anti-spoofing technology of GNSS is reviewed,and the purpose and significance of the research is given.Secondly,the feature extraction methods based on time-frequency analysis and bispectrum analysis are studied.Finally,the principles,characteristics,advantages and disadvantages of the commonly used time-frequency analysis methods are analysed,and then,aiming at its shortcoming,the bispectrum analysis method is introduced.In this dissertation,the definition,properties and disadvantages of bispectrum are presented.In view of the disadvantages of bispectrum such as large amount of computation,slow operation speed,etc.,two improving methods,integral bispectrum and diagonal sliced bispectrum,are studied.In the light of the nonlinear and non-stationary characteristics of GNSS signals,this thesis proposes two methods based on Wigner bispectrum to the spoofing identification of GNSS.The first spoofing identification method is based on the integral Wigner bispectrum and singular value decomposition,which makes use of integral to reduce the dimension of Wigner bispectrum,and uses the singular value decomposition to extract the feature.The second method is based on diagonal sliced Wigner bispectrum and feature fusion,in which,a variety of features of the diagonal sliced Wigner bispectrum are extracted and fused to form a feature vector.The experiments based on the simulated signals and real signals show that the two proposed algorithms can effectively identify the spoofing signal and achieve a good recognition performance in additive white Gaussian noise,and the performance of the second algorithms is better.