Research On GNSS Spoofing Interference Detection Technology Based On Machine Learning

The Global Navigation Satellite System(GNSS)civil signal is fragile due to the open structure,and spoofing interference poses a great threat to the security of GNSS-related services due to the strong concealment.Traditional detection methods are limited by models or need additional hardware,and the single detection parameter cannot fully and accurately reflect the characteristics of the spoofing signal.Dealing with the defects of the existing methods,this paper develops a software spoofing signal recognition system that can be widely used in civilian receivers,to realize intelligent recognition of spoofing interference.This paper provides technical support for the research and development of navigation spoofing detection technology and high-performance receivers,which is of great significance to maintaining the normal operation of the GNSS.The main contents of this paper are summarized as follows:(1)A semi-physical simulation platform for meaconing is built.Meaconing signals with different power and delay are generated by a GNSS signal simulator,an IF sampler is used to sample and store high-fidelity data,and a software receiver is used for acquisition,tracking and positioning.The effects of different meaconing power on the software receiver are compared and analyzed.(2)A meaconing detection method based on the Moving Variance(MV)of the carrier-to-noise ratio is designed,to solve the problem that the carrier-to-noise ratio detection method is easily affected by the satellite elevation angle.Firstly,set a sliding window to calculate the variance of the data subset in the window.Secondly,slide the window forward in a fixed interval and calculate the new data subset variance.Finally,set the threshold and detect the variance sequence.The experimental results prove the effectiveness of the proposed method.(3)A meaconing detection method based on improved Ratio is designed,to solve the problem that the traditional Ratio detection method ignores the distortion of the code loop Q-branch under the meaconing interference and depends on the carrier loop operating mode.Firstly,the I and Q correlation results of the same delay branch(early,prompt and late)are squared and then added to obtain the incoherent integration result.Secondly,the ratio of the early and late branches' sum to the prompt branch is calculated.Finally,the upper and lower thresholds are set to detect meaconing.The experimental results show that the detection rate of the improved Ratio is higher than that of the traditional Ratio.(4)A spoofing detection method based on Support Vector Machine(SVM)is designed.The characteristic quantities are extracted: Moving Average(MA)and MV of Signal Quality Monitoring(SQM),Early-Late Phase(ELP),the carrier-to-noise ratio MV and the receiver clock error change rate.The influence of different kernel functions on the detection performance is analyzed by TEXBAT(Texas Spoofing Test Battery)data set.The experimental results show that the accuracy of the coarse Gaussian radial basis kernel function is the best,reaching 92.31%.(5)A spoofing detection method based on Back Propagation(BP)Neural Network is designed.The effects of different activation functions on the detection performance are analyzed.The results show that the accuracy of Leaky Relu function is better than Sigmoid,Tanh and Relu functions,reaching 88.68%.The designed spoofing detection methods based on SVM and BP Neural Network are compared with the traditional SQM detection method.The experimental results prove the superiority of the proposed methods.The true positive rates reach 93.72% and 87.51% respectively,with the false positive rate set to 10%.