Shaping Distribution Identification of Probabilistically Shaped 64-QAM Signal Based on Signal Amplitude Characteristics

In this paper, we propose a new shaping distribution identification (SDI) method for the probabilistically shaped 64 quadrature amplitude modulation (PS-64QAM) signal based on signal amplitude characteristics. Firstly, the shaping distribution dependent features are obtained by calculating the amplitude mean, variance and the number of digitalized complex symbols within various ranges after amplitude normalized. Then, all these SD dependent features are extracted by a deep neural network (DNN) for achieving the identification of the shaping distribution information. The simulation results show that 100% SDI is successfully achieved with the entropy granularity as narrow as 0.1 bit/symbol and the OSNR lower than 18dB over back-to-back transmission. Similarly, 100% SDI has obtained as well over 800 km standard single mode fiber (SSMF) transmission, which indicates that the proposed technique is resilient towards fiber impairments.