Research On Unintentional Modulation Feature Extraction For Radar Emitter

The specific emitter identification(SEI) of radar is the development result of adapting to the new technology of radar and increasingly complex electromagnetic environment. It is the important development direction in the field of electronic warfare. The extraction of unintentional modulation feature is one of the key technologies for SEI system. unlike the basic characteristics of signal, it comes from the transmitter device’s non- linear characteristics, especially for the power amplifier. It has the characteristics of uniqueness, universal, independence, and scalability.This thesis focuses on the study of the extraction of the unintentional modulation features. Starting from the source of the unintentional modulation, and researching three algorithms, the main work is summarized as follows :1. Studied on the nonlinear modulation characteristics of transmitter, and analyses the principle. For the power amplifiers is the main source of unintentional modulation, a detailed analysis of its impact on the signal in the time domain and frequency domain have been made, and establishes mathematical models, to lay the foundation for subsequent research.2. For the poor performance of the anti- interference ability and discrimination about the special features. Researched the algorithm of extracting the unintentional modulation features based on the ambiguity function and wavelet transform. Firstly, It utilize the ambiguity function to map the signal to a two-dimensional space, and then, taking a wavelet transform on the oblique slices feature, finally, using the improved fisher LDA to select the best characteristics. The results of simulation and real data validate the effectiveness of the algorithm。3. For the higher order statistics has the advantages of restraining gauss noise and keeping the phase of signal, Researched the algorithm of extracting the unintentional modulation features based on higher order statistics. The algorithm has selected six features which associated with high order statistics to recognize the emitter. The results of simulation and real data validate the effectiveness of the algorithm.4. For the view of the time sequence is mapped from high-dimensional dynamical systems. Researched the algorithm of extracting the unintentional modulation features based on based on phase space and Gaussian mixture model(GMM). Reconstructing the phase space of the emitter signal and establishing the GMM for the centroid feature of the space. Then using the likelihood calculated by the model to do the classification. The results of simulation and real data validate the effectiveness of the algorithm. Meanwhile, does a comparative analysis for the three algorithms mentioned this paper.