英语文献总结

【1】Research on MPSK Modulation Classification of Underwater Acoustic Communication Signals
【2】Automatic Modulation Classification for Underwater Acoustic Communications

The purpose of Modulation Classification is demodulation non-cooperative signals，But 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.
This paper will introduce how to recognize the MPSK signals from a variety of communication signals and confirm the M value of PSK signals.
STEP1,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. However, the wavelet transform of MPSK is zero.
STEP2,the M value of PSK signals is confirmed by the feature parameter exploiting the four-order cumulant.

• step1

Firstly. The continuous wavelet transform of the signal r(t) is defined as formula1:

Secondly. We calculate wavelet transform of MPSK, MFSK and QAM respectively. Their mathematic model is shown as formula2:

After the above steps, we simulate the result of the continuous wavelet transform.

Finally .From this chart,It is obvious to identify the PSK signal from PSK, FSK, and QAM signals since the SNR is greater than 5 dB.

• step2

the three different ways of underwater communication modulation, MQAM, MFSK, MPSK can be distinguished by the means of continuous wavelet transform. Because the wavelet transform value of MPSK is zero, it is difficult to decide the M value of it. To solve this problem, we turn to the feature parameter of the fourth-order cumulants for help.
For a complex-valued stationary random process r(n), the fourth-order moments and cumulants can be defined in three ways. It is noted that the subscript of the cumulants represent various combinations of the signal and its conjugate function. For instance, the subscript 41 means that the cumulant uses three original siganls and one conjugating form. In order to calculate conveniently, we define a new variable Mpq, so the equations can be simplified as below.

Supposed the symbol E stands for the energy of the signal, the given table shows theoretical values of the fourth-order cumulants for MPSK signals. We need to construct classified feature vector F1,F2 to accomplish MPSK inner-class recognition, where F1 equals to the absolute value of C40 divided by C41, and F2 equals to the absolute value of C41 divided by C42. The values of the classified vector is varied with the order of PSK signals, which are expressed as this.

Next, look at the classification results. Using feature parameter F1, as shown in Figure 1, 8PSK signals are identified due to the distinct difference between it and other MPSK signals. Secondly, under the premise of ascertaining 8PSK signals, BPSK and QPSK signals are distinguished through the feature parameter F2 seen in Figure 2 .