Modulation Format Classification of Probabilistically Shaped M-QAM Signals Based on Nonlinear Power Transformation

In this paper, we proposed a modulation format classification scheme for probabilistically shaped quadrature amplitude modulation (PS-QAM) by extracting the amplitude information and nonlinear power spectrum after high-order Fast Fourier Transform (FFT). These modulation format dependent features include the signal’s amplitude mean and variance, the peak, variance and mean value of nonlinear power spectrum after 4
^th
, 8
^th
, and 16
^th
FFT, respectively. General category (modulation level) classification for PS-16QAM, PS32QAM, PS-64QAM and PS-128QAM with various shaping rate is successfully achieved. The simulation results show that 100% classification accuracy is successfully obtained with the OSNR lower than the normalized generalized mutual information (NGMI) threshold of 0.935 over back-to-back transmission. Similarly, 100% classification accuracy has obtained as well over single mode fiber (SMF) transmission, which indicates that the proposed scheme is resilient towards fiber impairments.