Research on MPSK modulation classification of underwater acoustic communication signals

On account of the complex underwater acoustic channel with severe attenuation, multipath effect, Doppler effect, as well as time and frequency spread characteristics, the task of inferring modulation of the underwater acoustic communication signals is extremely difficult and challenging. In this paper, we propose an automatic modulation recognition (AMR) scheme in order to recognize the MPSK signals from a variety of underwater acoustic communication (UAC) signals, considering the Gaussian white noise and multipath channels. The scheme deals with two parts: firstly, the feature extraction of UAC signals and the design of classifier. In the aspect of the feature extraction, a PSK modulation type is inferred using the amplitude of the variance for signals transformed by wavelet. This is necessary because the wavelet transformation of MFSK and QAM is a multi-step function and their variance amplitude of the wavelet transformation is greater than zero due to the performance of a multi-stage process. However, the wavelet transform of MPSK is zero. Secondly, the M value of PSK signals is confirmed by the feature parameter exploiting the four-order cumulant. This is necessary a more than two order cumulant can restrain Gaussian noise and has a good ability to adapt to signal to noise ratio (SNR). According to the proposed methods, the feature parameters with significant difference are obtained as the input of the classifier. Subsequently, the support vector machine (SVM) was employed as classifier for both inter-class and inner-class recognition. Both the train data and the test data to SVM were acquired by simulation, and we simulated the recognition rates of inter-class recognition and inner-class recognition respectively over the different training set, and we can anticipate that increasing the training data set improves the classifier performance. The experimental results show that the proposed scheme achieved the obvious effect of recognizing MPSK modulation signals remarkably and had an excellent recognition rate.