Research On Modulation Recognition Methods For Satellite Communications

Satellite communication has been widely used in civil and military scenarios because it is suitable for long-distance transmission and is not limited by terrain.Generally,an appropriate signal modulation format is selected according to the communication environment and user needs.With the development of satellite communication technology,the modulation formats used in satellite communication systems are becoming more and more diverse,and some modulation formats are difficult to distinguish,which makes the recognition of modulation formats in satellite communication systems more difficult.In this paper,satellite communication modulation formats will be recognized by extracting the characteristic parameters of the signal in multiple dimensions.In this paper,eleven satellite communication modulation formats{BPSK,QPSK,16QAM,8APSK,16APSK(4+12),16APSK(8+8),32APSK,64APSK(8+16+20+20),64APSK(16+16+16+16+16),128APSK,256APSK}are selected for the study.Starting with the traditional instantaneous features,higher-order cumulants features,constellation features and cyclostationarity features,the differences in satellite communication modulation formats under these features are first investigated.It was found that MAPSK modulation was difficult to recognize in both instantaneous and cyclostationarity features,that constellation features were more sensitive to noise,and that higher-order cumulants could only recognize 8APSK modulation and 256APSK modulation.Accordingly,this paper proposes an effective recognition method by analysing the phase characteristics of the modulation format:extracting the instantaneous phase of the received signal to construct a new signal,and using the higher-order moment spectrum characteristics of the new signal for recognition.Theoretical analysis shows that the noise variance of the original signal contains the amplitude value,which leads to a large variance,so the estimation convergence is poor,while the noise variance of the constructed signal is 1,which can achieve effective recognition.Based on the study of the modulation pattern characteristics,the second-order moment spectrum,eighth-order moment spectrum,twelfth-order moment spectrum,sixteenth-order moment spectrum and twentieth-order moment spectrum of the constructed signal and the|C₂₀|features?the|C₄₀|features?the|C₈₀|features?the=|₈₄|/|₆₃|~2 features are extracted analytically.Using the above nine feature parameters,the decision tree classifier and the support vector machine classifier were designed respectively to complete the modulation recognition technique for the eleven satellite communication modulation formats in this paper.Using the decision tree classifier,modulation recognition accuracy was 100%at data lengths of 2048 and signal-to-noise ratios higher than 10d B,and over 95%at signal-to-noise ratios of20d B and data lengths greater than 1000.Using the support vector machine classifier,the modulation recognition accuracy reached 100%at signal-to-noise ratios above10d B and data lengths of 2048,except for 128APSK modulation,which reached more than 95%at signal-to-noise ratios above 15d B.