| Recognition of Intra-pulse modulation is of great significance for pulse parameters estimation and radar emitter identification.Based on the background of the project,this paper firstly studies the recognition algorithms of typical radar intra-pulse modulated signals,and then studies the fast implementation of the algorithm on the hardware platform.Modulation recognition based on phase difference is a classical time-domain method,which is mainly used to classify simple modulation signals.And it has the advantages of simple principle,low complexity and easy realization.However,the existing algorithms require high SNR and can not adapt to the negative frequency signal,which is difficult to be applied in engineering.In this paper,two major improvements are proposed.First,dual-threshold phase unwrapping method is proposed to suppress the phase mutation caused by the noise and extend the adaptability to the negative frequency signals.Second,the classification process is optimized,and the preprocessing steps such as frequency-domain filtering are added to improve the performance under low SNR.The algorithm based on time-frequency images is mainly used to classify complex modulation signals such as polyphase coding and frequency coding.The main idea is to extract the characteristics of time-frequency images,and then use the machine learning methods for classification.This paper focuses on the LHOG feature improved by the HOG feature.The LHOG feature can represent the local shape information of the time-frequency image,and also has the translational invariance and rotational sensitivity,which is in line with the requirements of this subject.Finally,the performance of the algorithm is verified by the simulation experiments based on LHOG and SVM.In order to implement the algorithm on multi-core high-performance DSP platform,this paper designs the data stream structure,task planning and storage allocation in real-time processing system,and tests the system through real signals. |
PDF Full Text Request