Study On Modulation Recognition And Parameter Estimation Of Digital Communicate Signals

Automatic modulation identification and parameter estimation of digital communication signals are the key technologies in communication systems.This process is between signal detection and demodulation,and is of great significance in the fields of cooperative and non-cooperative communication such as radio spectrum management and electronic warfare.In recent years,with the rapid development of modern communication and signal processing technology,communication signal system and modulation style is becoming more and more complex and diverse,in addition,the signal transmission environment has become increasingly bad,all of these changes make the communication signal identification and parameters estimation requirements are getting higher and higher.According to the published literatures at home and abroad can be seen,the theory of automatic modulation recognition for communication signals is increasingly rich and perfect,especially in the low SNR of modulation recognition and parameter estimation problem under the condition of more concern.This paper mainly studies the automatic recognition technology of digital modulation signal,including feature parameters extraction,selection and classifier design and parameter estimation.In search of these parts,main work and innovation of this paper include the following aspects:Firstly,one popular feature extraction methods based on instantaneous information was researched and realized.A group of excellent feature parameter sets were proposed.However,the instantaneous information was easily affected by noise,so the wavelet threshold denoising algorithm is used to optimize it.Secondly,the design of classifier based on decision tree was studied.According to the extracted feature parameters,the classifier based on decision tree was designed,and the process of six kinds of recognition and classification of digital modulation signals was given.The classification performance of the classifier was simulated and analyzed.However,decision tree classifier is too dependent on each parameter and affected by the thresholds,so it cannot get good performance in low SNR conditions.Thirdly,to solve the shortcomings of the decision tree classifier,a support vector machine(SVM)classifier was designed.After analyzing the basic theory and classification principle,based on the same samples with the above-mentioned classifier,the classification performance was simulated with different SNR.In addition,analysis and comparison of two classifiers under the conditions of low SNR,SVM classifier can get better classification performance.Finally,analyzing and comparing the advantages and disadvantages of some different methods of the carrier frequency and symbol rate estimation.The improved algorithm is given,and the simulation results are optimized.Considering the unknown prior information,an improved joint modulation parameter estimation method based on cyclic spectrum and discrete spectral line extraction is proposed.The method utilized the relationship between modulation parameters and cyclic frequency,extracted the discrete spectrum line corresponding to the carrier frequency and symbol rate in the cyclic spectrum profile to estimate the parameters.Simulation results show that the detection ability of discrete spectral lines is improved effectively and the influence of truncation noise is overcome.