Research On Parameter Estimation And Blind Demodulation Technology Of CPM Signals

Nowadays,wireless spectrum resources are increasingly tight,so it is urgent to find a modulation method with high spectral efficiency.Continuous phase modulation(CPM)technology is widely used in digital communications due to its excellent spectrum utilization and power utilization.In CPM,the phase changes continuously over time,rather than taking specific discrete values at specific time intervals as in other digital modulation techniques.With the continuous development of technology,CPM technology has gradually become one of the important technologies in the field of digital communication.In the non-cooperative communication system,as an unauthorized third party,The signal sent by the transmitter and its related parameters are unknown to the receiver.Therefore,in order to obtain the data information sent by the sender,the receiver needs to process the received signal.In order to recover the information of unknown signals,it is necessary to perform parameter estimation and demodulation of the received signal.In digital communication system,accurate estimation of modulation parameters is very important for realizing high speed and reliable data transmission.Therefore,the main research content of this paper is the parameter estimation and blind demodulation technology of CPM signals.Firstly,this paper introduces the classification of CPM signals and their basic concepts,and expounds the basic expressions and phase information of CPM signals,analyzes its spectrum,and describes the decomposition of CPM signals,which provides a theoretical basis for the subsequent demodulation of CPM signals.Secondly,the modulation parameters of the CPM signal are estimated.For the modulation order,a modulation order estimation algorithm based on cyclic spectrum is adopted,which estimates the modulation order by observing the number of discrete spectral line roots in the cyclic spectrum of the signal.For the correlation length,a correlation length estimation algorithm based on autocorrelation characteristics is adopted,which estimates the correlation length by observing the average autocorrelation function of the CPM signal,then analyzing its characteristics,and then setting a specific threshold.For pulse type,the original algorithm is based on the pulse type recognition algorithm of the support vector machine,which calculates the variance,kurtosis and slope of the CPM signal,and then inputs these three features to the support vector machine for prediction.In this paper,an algorithm for identifying pulse types using convolutional neural networks is proposed,which has a high accuracy rate.For the modulation index,the traditional algorithm is to define a cost function,and estimate the modulation index by searching for the coordinates of the location where the maximum value of the cost function is located,but the algorithm has certain limitations,this paper proposes an improved modulation index estimation algorithm,which is suitable for both Single-h CPM and Multi-h CPM signals,and can accurately estimate the modulation index when the signal-tonoise ratio is 13 d B,and also increases the range of signal estimation.Finally,the CPM signal is blindly demodulated.Firstly,the maximum likelihood sequence detection algorithm based on Viterbi is described,and on this basis,a blind demodulation algorithm based on CPM signal decomposition is studied,which is divided into two types,one is optimal detection and the other is suboptimal detection.The blind demodulation algorithm based on CPM signal decomposition can effectively reduce the demodulation complexity of the receiving end.