Research On Jamming Detection And Recognition Technology In Satellite Broadband Frequency Hopping Systems

Satellite broadband frequency hopping(FH)systems have the performance advantages of anti-fading,anti-interference and low interception.However,the ever-increasing jamming equipment and detection equipment are weakening the anti-jamming gain from broadband high-speed FH.In order to ensure the transmission reliability and resource utilization of the communication system,satellite broadband FH systems need to have the ability to detect and identify typical man-made malicious jamming to adjust the anti-jamming strategy in time.For this reason,this thesis mainly studies the jamming detection and classification technology of typical man-made malicious jamming in satellite broadband FH systems.The main content of this thesis consists of four parts:The second chapter designs the jamming detection and recognition scheme for the satellite broadband FH system.After the interference threats faced by the satellite broadband FH system being analyzed,a jamming detection and recognition scheme based on orthogonal subband segmentation is proposed and the wide-band jamming cognition is transformed into multiple narrow-band jamming cognition,which effectively reduces the implementation complexity of the scheme while ensuring the accuracy of frequency domain analysis.Next,a comprehensive strategy for multi-subband jamming cognition is proposed.By combining the jamming cognition results in multiple subbands,the wide-band jamming cognition results are obtained.The third chapter studies the jamming detection algorithm in the satellite broadband FH system.According to the difference in time-frequency characteristics between the FH signal and the jamming signal,a jamming detection algorithm based on Multi-segment Signals Spectrum Clustering(MSSC)is proposed.First,the signal cluster detection algorithm based on Welch spectrum and Forward Consecutive Mean Excision(FCME)is explained and the detection threshold factor is theoretically analyzed.Then the jamming cluster identification algorithm and the silent-period signal segment searching algorithm both based on MSSC are proposed.Finally,the complexity of the algorithm is analyzed.The simulation results show that when the signal-to-noise ratio(SNR)of the FH signal is greater than-4d B,the jamming false detection probability of the traditional algorithm is greater than 60%while the MSSC algorithm is less than 10^-4,and the estimation error of the jamming parameters by MMSC is significantly reduced.The fourth chapter studies the jamming classification algorithm in the satellite broadband FH systems during the silent period.First,the jamming classification algorithm based on Manual Feature Extraction and Deep Neural Network(MFE-DNN)is proposed.A set of excellent multi-domain jamming features is extracted and a jamming classifier based on DNN is trained.The simulation results show that the MFE-DNN algorithm can accurately classify 12 jamming patterns when the Jamming-to-Noise Ratio(JNR)reaches-4d B.Then,the performance of jamming classifiers based on automatic feature extraction with convolutional neural network(CNN)under different input data objects are studied,and a jamming classification algorithm based on Convolutional Neural Network-based Joint Multi-Domain Feature Extraction(CNN-JMDFE)is proposed.The simulation results show that when JNR reaches-6d B,the CNN-JMDFE-based jamming classifier can accurately classify 13jamming patterns,which is significantly better than the performance of CNN-based jamming classifier that only inputs a single data object.And compared with the MFE-DNN jamming classifier,CNN-JMDFE-based jamming classifier can significantly improve the jamming classification accuracy of 11 types of jamming patterns at low JNR,with a gain of 2d B?12d B.Chapter 5 designs a satellite broadband FH jamming cognition system model based on MSSC and CNN-JMDFE algorithms.The simulation results show that the model has a good classification effect on both single jamming and compound jamming.The classification performance of 14 types of single jamming in the non-silent period is consistent with that in the silent period,and the 10 types of compound jamming can be accurately classified when JNR reaches 0d B.Compared with the CNN-JMDFE-based jamming classifier without jamming detection,this model has a performance gain of6d B and 8d B in classification accuracy of single jamming and compound jamming,respectively.As for the wide band,this model can accurately estimate jamming parameters and correctly determine the type of jamming.