Feature Classification And Recognition Of Radar Pulse Modulation Based On Wavelet Transform

Automatic modulation classification is to determine the modulation type of the received signal without a priori information.It is an indispensable important link between signal monitoring and demodulation.It is widely used in military and civilian communications fields,such as cognitive radio,Adapt to modulation and coding,spectrum monitoring and electronic warfare systems,etc.Especially in electronic warfare,the algorithms related to the modulation recognition of radar pulse signals combined with the real-time recognition of the modulation type by the receiver equipment are of great significance on the battlefield.Obtaining the radar's own system,working status and other information through the modulation type can fully occupy the initiative.However,the current electromagnetic environment is becoming more and more complex,and the anti-interference of the recognition algorithm needs to be strengthened;various broadband and ultra-wideband radar signals also bring great challenges to the processing speed;the types of intra-pulse signal modulation are increasing;these problems are all related to the intra-pulse signal recognition presents new challenges.This thesis combines the existing wavelet and wavelet packet feature extraction algorithms,introduces cluster centers to calculate the distance between centers to select different feature extraction methods,and studies the wavelet base and classifier categories,and designs a feature extraction method.And implemented on the basis of hardware for "complex electromagnetic environment monitoring".It mainly includes the following contents.Firstly,theoretical research and analysis of existing feature extraction methods.Research commonly used radar feature extraction methods,the basic knowledge of wavelet theory,and analyze the advantages of wavelet theory for extracting radar pulse features.Secondly,an improved feature extraction method combining wavelet transform and wavelet packet transform is proposed.By comparing various methods of wavelet analysis,including single wavelet decomposition and reconstruction of discrete signals,wavelet packet decomposition and reconstruction of discrete signals.At the same time,it is also necessary to select an appropriate wavelet base and decompose the appropriate number of layers based on the characteristics of the pulse signal itself in the frequency domain.Improve the existing wavelet feature extraction algorithm and compare with the existing feature extraction method.Finally,the simulation verifies the performance superiority of the improved modulation feature recognition method.The improved method is compared with the existing feature extraction method based on wavelet theory in noise immunity and parameter transformation stability to analyze the improvement of performance.Finally,the algorithm used is implemented in C ++ and added to the existing software project,so that the monitoring function of the original sampler device is expanded.A set of theoretical system and hardware realization equipment with complete recognition of radar pulse modulation signal characteristics based on wavelet transform are established.The simulation results of the algorithm show that the entropy feature extracted by the combination of wavelet and wavelet packet has better classification accuracy than the existing classification methods.And apply it to the existing hardware equipment for testing,it can indeed achieve automatic modulation type classification,although the parameters such as pulse width and carrier frequency have a certain effect on the characteristic parameters,but the stability of the algorithm when these parameters change It can extract the characteristic parameters with better classification effect,and provides a potential solution for automatic classification counting.