Spoofing Detection Based On Channel Difference And Decision Fusion

GNSS spoofing is an intentional jamming,which is very similar to a true navigation signal.Spoofing interference deceives a GPS receiver to capture the jamming signal,which may cause serious security problems,such as erroneous synchronization time and false position.Compared with the blanket jamming,spoofing jamming has better invisibility,simpler device,and it is harder to be detected.Spoofing jamming is becoming one of the main threats to satellite navigation systems.It is necessary to develop an effective method for satellite navigation spoofing detection on the basis of the existing technology.This paper implements spoofing detection and classification from two aspects: statistical learning theory and machine learning theory.Existing works ignore the statistics difference between the satellite channel and the spoofing channel,and the decision fusion method has not been used to detect GNSS spoofing.This dissertation aims at compensating the foregoing research problems.The main research works are as follows:1.The part of introduction presents the research background and significance of GNSS anti-interference,the research status of anti-interference,the existing spoofing detection algorithm and jamming suppression algorithm,the research status and application of decision fusion algorithm,and the advantages of the decision fusion method.2.A spoofing detection method based on channel difference from statistical theory aspect is proposed.Firstly,we introduce the process of spoofing detection,the probability distribution models of satellite communication,and Lutz model of satellite channel.Secondly,the principles of Kolmogorov-Smirnov(KS)detection,Cramer-von Mises(CVM)detection and Anderson-Darling(AD)detection of goodness-of-fit test are studied.Finally,the spoofing detection method is proposed based on the goodness-of-fit test under the two states of satellite channel.The effectiveness of the method is verified by simulation experiments.3.The theory of decision fusion is elaborated.First of all,the method of multiple classifiers combination: serial combination,parallel combination,and cascade combination are introduced.Then,the three rules: AND rule,OR rule,and K/N rule of cognitive radio spectrum sensing are presented.Finally,the optimal fusion of K and N are deduced based on the rule of minimum bayes risk and the rule of minimum error probability.4.A spoofing detection method based on decision fusion from machine learning theory aspect is proposed.Firstly,we extract the singular values of the wavelet transformation coefficients of the real signal and spoofing signal as a feature vector.Secondly,based on the same feature vector,three different classifiers,i.e.SVM,PNN and DT,are adopted to the identification of the spoofing.Thirdly,with the recognition result of each classifier,final identification result is obtained by decision fusion with the K-out-of-N rule.Simulation results demonstrate the decision fusion method can make full use of different classifiers,and improve the overall recognition rate and system performance.