Modulation Recognition And Parameter Estimation Algorithm Of Frequency-modulated Signal

Blind processing of communication signals is an important area of communication signal processing.Blind Modulation Recognition(BMC)and parameter estimation are widely used in spectrum management and non cooperative communication of monitoring,cognitive radio.Blind Modulation Recognition and parameter estimation are the precondition for obtaining information,which is the basis for information demodulation in receivers.Blind processing does not contain any additional information and pilot sequences.To realize blind modulation recognition and parameter estimation,it is necessary to extract the characteristics of the received signal in time domain and various transform domains,and design the recognition scheme based on this.With the increasing tension of spectrum resources,the situation of receiving multiple signals from single antenna is becoming more and more popular.The received signals generally overlap in time and frequency domain,and the blind separation of time-frequency mixed signals has become an urgent problem to be solved.Among them,signal modulation recognition and parameter estimation are the basis of blind signal separation.This paper mainly studies the BMC of frequency modulation signals and the blind separation of frequency aliasing signals.The main contents and contributions are as follows:1.Combined with the characteristics of multiple transform domains of FM signals,a modulation recognition algorithm suitable for various FM signals is proposed.The main FM signals include radar FM signals(LFM,HFM,SFM),continuous phase MFSK signals(CPM-2FSK,CPM-4FSK),discontinuous phase MFSK signals(2FSK,4FSK)and noise FM signals(noise-FM).The main characteristic parameters used include the power spectrum discrete line number,the instantaneous autocorrelation peak number,the normalized instantaneous amplitude spectrum maximum and the normalized instantaneous frequency spectrum maximum value,and the instantaneous normalized instantaneous frequency mean parameter is used to realize the intra class discrimination of continuous phase MFSK signals.Based on the cyclostationarity of signals,the parameter estimation algorithm of SFM signals is improved.The parameters of the SFM signal are estimated by using the cyclic cumulant section,and the estimation accuracy is improved by combiningCarson formula.2.Based on the cyclostationarity of signals,the separation algorithm of time frequency aliasing signals is designed.The algorithm mainly utilizes the difference between the the non-zero cyclic frequency of two cycle cumulants,the two order cyclic conjugate cumulants and the the four order cyclic cumulants.Based on cyclic cumulants and cyclic spectrum characteristics,the parameter estimation of time frequency overlapped signals is studied.There is a certain relationship between the cyclic cumulants,the cyclic spectrum's cross sections and the characteristic parameters of the overlapped signals.This paper improves the estimation accuracy combined with the spectrum thinning algorithm.Finally,the blind separation algorithm of digital signals is explored by using the sideband characteristics of signals.The side band characteristics of the signal are innovatively utilized to achieve the separation of spectrum and achieve better separation performance for mixed signals with spectral overlap less than 50%.