Recongnition And Parameter Estimation Of Hybrid Modulation Signal In TT&C

Tracking Telemetry and Command System(TT&C)has been widely used in military and civil fields,which is not only an important link between spacecrafts and ground base station for transmission of various type of information but also an useful means for tracking and positioning.Hybrid modulation is an important type of modulation in TT&C,so it becomes very necessary to study the classification and recognition of hybrid modulation signal.Therefore,both from the perspective of theory and application,the parameter estimation and recognition of hybrid modulation signals have very important academic significance and military application value.In this context,this thesis mainly takes hybrid modulation signal in TT&C as the research object,and the research content mainly includes parameter estimation,de-noising of hybrid modulation signal and modulation recognition.Firstly,the corresponding parameters of hybrid modulation signal are estimated.Parameter estimation and modulation identification are inseparable.On one hand,both of them are important contents of communication reconnaissance.On the other hand,they are closely related.Parameter estimation provides the basis for subsequent modulation identification.In this thesis,the signal-to-noise ratio(SNR),carrier frequency,symbol rate and modulation index are estimated.Among them,SNR estimation mainly adopts the method of second-order fourth-order moment(M2M4)and subspace decomposition.The carrier frequency estimation is mainly based on the relevant information of signal frequency domain.The symbol rate estimation is carried out by using the method of quadratic wavelet transform.The modulation index estimation is performed according to the time-frequency structure of the signal.Secondly,de-noising of hybrid modulation signal is introduced.The de-noising of the hybrid modulation signal is a pre-processing procedure for classification and recognition of hybrid modulation signal,which can improve the accuracy and reliability of modulation recognition.In order to realize blind de-noising and reduce the need for prior information,this thesis selects the wavelet threshold de-noising method.The influence of different wavelet bases,scale factors and threshold functions on the de-noising performance is analyzed.Forthermore,the influence of denoising algorithm on parameter estimation performance is studied as well.Finally,recognition of hybrid modulation is introduced.This thesis analyzes the characteristics of various hybrid modulation signals,including AM-FM,FM-FM,PCM-FM,PCM-PM,PCM-BPSK-PM,PCM-QPSK-PM,PCM-FSK-PM.Six characteristic parameters to identify these signals are selected.For signal classification,this thesis mainly introduces four classifiers: decision tree,BP neural network,support vector machine(SVM)and vector matching.Among them,the decision tree carries out classification in a serial manner,which needs to determine a set of thresholds in advance,while BP neural network,SVM and vector matching are all used for classification ccording to the unified characteristics of signal,which doesn't need to determine the decision threshold of signal in advance.Howerer,these approaches require a large number of training samples and have great computational complexity.