Deceptive Jamming Detection Of Satellite Navigation Based On ELM

Global navigation satellite system has been widely used in national defense construction and daily production and life.The safety of navigation system has attracted more and more attention of researchers.In particular,the increasingly complex electromagnetic environment also makes researchers pay more attention to the reliability of satellite navigation system instead of blindly pursuing navigation and positioning accuracy.The signal transmitted by deception equipment is extremely "cunning".It disguises as a real signal,which greatly increases the difficulty of successfully detecting malicious interference and makes the traditional detection methods difficult to apply.In recent years,with machine learning shining in more and more application fields,the introduction of machine learning algorithm into deception and interference detection has attracted more and more attention.Considering that extreme learning machine(ELM)has the advantages of simple structure,strong generalization ability and fast learning speed,ELM and its deformation are introduced to detect the existence of deceptive signals.Specifically,this paper studies the deception and interference detection of satellite navigation from the following two aspects.Firstly,a deceptive interference detection method for satellite navigation based on ELM model and particle swarm optimization(PSO)algorithm is proposed.This method uses ELM model to learn and classify the processed data.Considering that the input weight matrix and hidden layer bias matrix in the initialization process of ELM model are randomly assigned by the system,which brings some uncertainty to the final detection results.Compared with other modern optimization algorithms,PSO algorithm has the advantages of simple parameter setting,clear principle and strong performance.Therefore,the PSO algorithm is used to optimize the two parameters randomly set during the initialization of the ELM model.Experimental results show that this method has good applicability in detecting deceptive interference.Secondly,considering that in the actual scene,the proportion of real signals is far greater than that of deceptive signals,that is,there is an imbalance in the collected data.To this end,before feeding the training set to the classification model,a synthetic minority class oversampling technique is used to oversample the small number of classes,so that the reconstructed dataset is relatively balanced.At the same time,aiming at the disadvantage that ELM model is unable to deal with batch data,its deformation is introduced,that is,online sequential extreme learning machine model,and the model is fine tuned by updating the formula.Experimental results show that this method still has good performance in detecting deceptive interference under unbalanced data sets.