Mitigation of distortions in fiber-optic communication systems using optical neural network-based equalizer

A feedforward optical neural network (ONN) based on Silicon photonics nanocircuits is proposed for the equalization of linear and nonlinear impairments of fiber optic communication systems. The activation function is realized using a semiconductor saturable absorber mirror (SESAM) or a highly nonlinear waveguide (HNLW). The proposed ONN can be placed at the fiber optic link output just before the coherent receiver. The numerical simulation of a 28 GBaud quadrature phase shift keying (QPSK) signal over a long haul dispersion-managed fiber optic link showed that the performance can be significantly improved by using the ONN based on HNLW. The dominant impairment in the long haul dispersion managed link is the fiber nonlinearity. To illustrate that the proposed ONN can mitigate fiber dispersion, simulations of a short-haul fiber optic link consisting of a standard single mode fiber (SSMF) is carried out and results showed that the ONN based on SESAM is quite effective in mitigating the dispersive impairments of the fiber optic system.