Research On Modulation Classification Technique For Underwater Acousitc Communication Signals

The automatic classfication of modulated signals is an important research field in modern communication technology.With the exploitation of ocean and ocean as important military strategy,the research on the modulation classification of underwater acoustic communication signals has been increasing rapidly.However,underwater modulation classification is confronted with many challenges,such as complex channel with strong multipath,narrow bandwidth,long delay,Doppler effect and underwater impulsive noise.Firstly,the underwater acoustic channel model and the impulsive noise model are established.The underwater acoustic channel model is combination of the physical model and the statistical model.The physical model is based on the BELLHOP simulation software with inputting underwater environmental measured data,and the statistical model is time varying multipath Ruili fading model.Impulsive noise model is established on the basis of the Alpha stable distribution theory.Meanwhile,the general SNR is defined as the physical parameter to measure the noise intensity.The established models provide a platform and simulation environment to simulate the performance of subsequent classification algorithms.Secondly,a method of signal modulation classification based on the fractal dimension of frequency domain is proposed for underwater impulsive noise.Due to the frequency spectrum of the signals are less affected by the impulse noise with1(27)? ?2,the box dimension features of the frequency domain are extracted,and the stability of the feature is simulated under the impact noise of different parameters.In order to further enhance the performance of the algorithm,a non-linear transform pretreatment method of impulse noise is utilized.The simulation results show that the algorithm can effectively classify SSB,4FSK,QPSK and OFDM signals under impulsive noise,and the performance and stability are improved significantly after pre-processing.Thirdly,for the time-varying underwater acoustic multipath channel,a signal modulation classification algorithm based on the spectral energy entropy of Stockwell transform is proposed.In order to measure the difference of the spectral energy distribution of different signals,the energy entropy of the spectrum is extracted as the classification feature.The simulation results show that the method can effectively classify SSB,FM,4FSK,QPSK and OFDM signals in underwater acoustic channel,and after adding impulse noise in channel,the algorithm still has better performance.Finally,on the basis of the single feature extraction method above,the feature of time-domain envelope sample entropy is added,and signal modulation classification is carried out based on multi-feature fusion methods.Combining with the signal features of time domain,frequency domain and S-transform domain,reaserch two modulation recognition methods of multi-feature fusion.One method is based on DS evidence theory,with a basic probability assignment strategy based on Mahalanobis distance and exponential probability distribution function.The fusion of DS evidence theory and decision criteria are used to realize the modulation classification of signals.Another one is based on SVM with multi-featrues vector.The simulation results show that the two methods based on multi-features fusion have better performance under underwater acoustic channel and impulsive noise environment.Compared with the single feature method,the performance of them is improved significantly,and the classificable signal type is increased,but the computational complexity increases significantly,so in practice it need to compromise.In addition,based on the multi-feature fusion method,the MC-MFSK and OFDM signals are distinguished better.