QAM Modulation Classification Based on Düffing Oscillator

A new modulation classification algorithm based on Düffing oscillator has been proposed in this paper. It is known that the Düffing oscillator is not sensitive to noise, but resonant to periodic signals, even if ratio of signal to noise is very low. Thus it is applied for modulated signal classification, and an algorithm for QAM modulation classification is put forwards. In the algorithm, a Düffing oscillator is first used for depressing noise level of the input modulated signal. Which is a M-ray QAM signal with noise. Then their features related to its amplitude are extracted from the output signal. These features are clustered into several groups in phase space, and the order M is determined by the number of clusters. Simulating results covering 8QAM, 16QAM, 32QAM and 64QAM modulation schemes show that this algorithm has good performance in the presence of noise.