Parallel Bisection-based Distribution Matching for Nonlinearity-tolerant Probabilistic Shaping in Coherent Optical Communication Systems
Mengfan Fu, Qiaoya Liu, Huazhi Lun, Hexun Jiang, Yiwen Wu, Xiaomin Liu, Zhiyuan Yang, Lilin Yi, Weisheng Hu, Qunbi Zhuge
Abstract
A nonlinearity-tolerant constant composition distribution matching (CCDM) scheme based on a parallel bisection architecture is proposed. Compared with CCDM based on arithmetic coding (AC), the bisection-based CCDM (BS-CCDM) can achieve the same linear performance and an improved tolerance to fiber nonlinear impairments in unrepeated fiber transmissions. The same linear performance of AC-CCDM and BS-CCDM is evaluated both in simulations and experiments with various modulation formats and shaping rates. The improved fiber nonlinear performance is demonstrated in simulations with probabilistically shaped dual-polarization 16-ary quadrature amplitude modulation (DP-16QAM) signals for unrepeated transmission over 250 km of standard single-mode fiber (SSMF). BS-CCDM provides a ∼0.22 dB gain in effective signal-to-noise ratio (SNR) on average over AC-CCDM for a variety of spectral efficiencies with a block length of 64. And a gain in achievable information rate (AIR) of ∼0.11 bits/4D-symbol is obtained correspondingly. In addition, by further optimizing the look-up tables (LUTs) in BS-CCDM, a better trade-off between the linear performance and nonlinearity tolerance is achieved. Compared with AC-CCDM, the optimized BS-CCDM provides the gain of 0.48 dB and 0.7 dB in effective SNR for the block length of 64 and 128, respectively.