Research On Signal Detection And Identification Technology Of GNSS Interference Source Based On Data Driven

Services of position,velocity and time provided by global navigation satellite system(GNSS)have the characteristics of high precision,all-weather etc.GNSS are widely used in all aspects of life.Therefore,the security of its services becomes more and more important.However,Satellite signals become extremely weak when they reach the earth,so they are vulnerable to interference.Quick detection and identification of interference source signals that may be encountered in satellite navigation signals is of great significance for the construction of the whole interference monitoring system and improving the anti-interference performance of navigation system.Firstly,global positioning system(GPS)in GNSS has been introduced.Meanwhile,the frequency characteristics of GPS signal have been analyzed as well.Then,types of GNSS interference source signals have been analyzed.After that,some typical interference signals were modeled.Then,followed by mathematical basis of the models of interference signals,the original data of interference signals have been obtained through the closed-loop data acquisition system.Secondly,traditional technologies of interference detection have been studied.The energy detection algorithm based on time-domain analysis,the consecutive mean excision(CME)and forward consecutive mean excision(FCME)algorithms based on frequency-domain analysis were transferred to complete the detection of GNSS interference source signals.The simulation results show that all three algorithms of interference detection can detect the interference signals.But the detection performance is subject to the condition of jamming-to-noise ratio(JNR).If the JNR is lower than-5d B,the detection probability of energy detection algorithm will decrease sharply;if the JNR is lower than-8d B and-12 d B respectively,the detection probability of CME and FCME algorithms will be reduced respectively to varying extent.Finally,the data-driven method,that is,the deep neural network algorithm was used to detect and identify the GNSS interference source signals.The used types of deep neural networks include convolutional neural network(CNN)and residual network(Res Net).And then the above networks were trained,tested and verified.In terms of detection of GNSS interference source signals,CNN was used to detect the interference signals.The results show that when the JNR is greater than-17 d B,the network can accurately detect the interference signal;In terms of identification of GNSS interference source signals,CNN and Res Net were used to identify the interference signals respectively.The results show that under the conditions of multiple types of interference signals(12 types)and low JNR(-4d B),the identification accuracy of CNN is only 72%,while the identification accuracy of Res Net is12% higher than that of CNN.If the method proposed in this thesis can be applied to engineering practice,it will have certain practical significance to improving the detection and identification performance of interference source signals.