Open AccessBlind time domain nonlinear compensator embedded in the constant modulus algorithm
Author(s) -
Junhe Zhou,
Yunwang Zhang
Publication year - 2019
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.27.022794
Subject(s) - nonlinear system , control theory (sociology) , computer science , fast fourier transform , time domain , algorithm , frequency domain , signal processing , computational complexity theory , iterative method , nonlinear distortion , compensation (psychology) , digital signal processing , bandwidth (computing) , telecommunications , artificial intelligence , psychology , amplifier , physics , control (management) , quantum mechanics , psychoanalysis , computer hardware , computer vision
In this paper, a blind nonlinear compensator is proposed for the compensation of fiber nonlinearity. The nonlinear compensator is embedded in the constant modulus algorithm (CMA) which consists of two stages. In the first stage, the linear filter coefficients are calculated while the nonlinear CMA is performed in the second stage. After the nonlinear coefficients reach the stable state, the two-stage CMA switches to the conventional CMA followed by a fixed time domain nonlinear equalizer because the fiber nonlinearity within the transmission link does not vary as the polarization rotation does. This greatly reduces the digital signal processing (DSP) complexity and saves the computational efforts. In comparison with the existing nonlinear compensation techniques such as the back-propagation (BP) method and the Volterra series based nonlinear equalizer (VSNE), the proposed method does not need fast Fourier transform (FFT) or iterative procedures, which significantly reduces the computational complexity. Since the method is performed blindly, no prior information about the transmission link is required, which greatly facilitates the implementation of the nonlinear compensation technique. In addition to these, the coefficients of the nonlinear compensator are optimized adaptively, and it outperforms the existing methods. Finally, the method can work in conjunction with the existing nonlinear compensators, such as the BP method, and significant performance improvement is found over the original nonlinear compensator.