Research On The Design Algorithm Of Interference Signal For Anti-intelligent Recognition

With the rapid development of modern electronic information technology,the recognition technology of radar jamming signal based on neural network has become more and more mature.In order to obtain a better jamming effect,it is necessary to consider how to avoid being identified by the enemy’s electronic reconnaissance system when designing the radar jamming signal.However,the current radar jamming signal is difficult to deal with the intelligent identification based on neural network,and it is difficult to adapt to the rapid development of modern electronic warfare technology.Therefore,an anti-intelligent recognition radar jamming signal generation method is urgently needed.this thesisAiming at the fact that neural network is more sensitive to input data,this thesis realizes the generation method of anti-intelligent identification interference signal through neural network confrontation attack.According to the structure and parameters of the neural network for intelligent identification of enemy radar jamming signals,the anti-intelligent identification jamming signal generation methods based on white box attack and black box attack are respectively implemented,which can greatly reduce the identification accuracy of the enemy electronic reconnaissance system.Rate.The main work of this thesis is as follows:1.Seven datasets of radar jamming signals are constructed.The generation principles of seven kinds of radar jamming signals are analyzed in detail,and the radar jamming signal datasets with various interference-to-noise ratios are generated,which provide data support for the radar signal intelligent identification model and the generation method of anti-intelligent identification jamming signals.2.Three kinds of radar jamming signal intelligent recognition models are built and trained.On the basis of the three radar jamming signal intelligent recognition models,the neural network model suitable for the data set in this paper is obtained by modifying the parameters,all of which can achieve a better recognition accuracy,and provide model support for the generation method of anti-intelligent identification jamming signals.3.When the structure and parameters of the neural network for intelligent identification of enemy radar jamming signals are known,an anti-intelligent identification jamming signal generation method based on white-box attack is studied.According to the different optimization objectives,the generation methods of antiintelligent identification interference signal based on direct optimization,gradient optimization and boundary decision analysis are implemented respectively.The method can obviously reduce the recognition accuracy of the intelligent recognition model of the enemy radar jamming signal and improve the anti-recognition performance of the jamming signal under the condition of adding a small disturbance.4.In the case of unknown enemy radar jamming signal intelligent identification neural network structure and parameters,the anti-intelligent identification jamming signal generation method based on black box attack is studied.According to the difference of the direction search basis,the standard orthonormal basis and the DCT orthonormal basis are respectively used for direction search,and the anti-intelligent identification interference signal generation method based on Decision-based black box attack is realized.The method can realize the generation of anti-intelligent identification interference signals under the condition of less query times.The method implemented in this paper has been verified by simulation.The results show that the anti-intelligent identification jamming signals generated by these methods can effectively reduce the recognition accuracy of the intelligent identification model of enemy radar jamming signals.