Research On MPPSK-modem Based On SAE Deep Learning Network

The radio spectrum resource is an important strategic resource.With the continuous development of information socity,the demand for wireless communication continues to increase,so that the original limited spectrum resoures becomes increasely scarce.Thus,it is particularly important to improve the efficient use of the spectrum resoures.Therefore,this paper discussed efficient MPPSK(M-ary Phase Position Shift Keying)modulation,which has ultra narrow spectral characteristics,in detail in different channel environments and different modulation parameters,and introduces the new auto encoder based on stack(SAE:Stacked Auto Encoder)demodulation method of deep learning network,through compared with the traditional demodulation method,constantly adjust the network parameters,and thus enhance the demodulation performance,which has important theoretical significance and practical value.First of all,this paper introduces the modulation theory and spectrum characteristics of EBPSK and MPPSK with high efficiency.Taking MPPSK as an example,this paper expounds the key technologies of high efficiency modulation signal demodulation and the principle,characteristics and design points of the shock filter;Secondly,this paper introduces the SAE deep learning network and Softmax regression model,through simulation,find the optimal network model of SAE deep learning and deep learning,SAE network and Softmax regression model for multi classification problems.To compare the signal detection efficiency among traditional methods and SAE deep learning network method,the EBR of complex modulation system is simulated.Finally,using MPPSK modulation as the research object,analyzes the advantages and disadvantages of various traditional judgment method and detection difficults of MPPSK signal,then the SAE deep learning network decision method is introduced into the efficient communication system,and in different channel environments and different modulation parameters,the advantages and disadvantages of the traditional decision scheme and SAE deep learning network method are compared.The simulation results show that under the selected parameters in this paper,the method based on the MPPSK demodulation performance of SAE deep learning network is excellent.