Open AccessBlind back-propagation method for fiber nonlinearity compensation with low computational complexity and high performance
Author(s) -
Junhe Zhou,
Yuheng Wang,
Yunwang Zhang
Publication year - 2020
Publication title -
optics express
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.394
H-Index - 271
ISSN - 1094-4087
DOI - 10.1364/oe.387572
Subject(s) - computational complexity theory , nonlinear system , compensation (psychology) , channel (broadcasting) , computer science , optics , algorithm , reduction (mathematics) , dispersion (optics) , span (engineering) , bit error rate , transmission (telecommunications) , control theory (sociology) , mathematics , telecommunications , physics , artificial intelligence , psychology , geometry , civil engineering , control (management) , quantum mechanics , psychoanalysis , engineering
In this paper, a blind back-propagation (BP) method for fiber nonlinearity compensation with low computational complexity and high performance is proposed. The BP method compensates the fiber chromatic dispersion step by step. Between two linear steps, the proposed method compensates the fiber nonlinearity with the nonlinear tap coefficients optimized by the nonlinear least square method (NLSM). Unlike the traditional BP method, the proposed method takes into account the SPM, the intra-channel XPM and the intra-channel FWM effects while it is purely blind and requires no prior information of the transmission link except the total accumulated chromatic dispersion, e.g., the BP step in the proposed algorithm can be set as an arbitrary value which has no connection to the physical span length. The computational complexity of the proposed method is much lower (less than 50%) than the conventional BP method with one step per span, because of the reduction of the total number of steps. Meanwhile, the method improves the nonlinearity compensation performance in comparison to the standard BP method with one step per span at the optimal input power while maintaining the same computational complexity.