Advanced nonlinear equalizer for Filter Bank Multicarrier-based Long Reach-Passive Optical Network system

The performance of the intensity-modulated Filter Bank Multicarrier (FBMC) system using direct detection with advanced nonlinear equalizer in Long Reach-Passive Optical Network (LR-PON) is presented in this paper. First, the performance of the FBMC system and its nonlinearity in the channel are analyzed. We introduce two nonlinear equalizers, namely, Artificial Neural Networks-Nonlinear Feed-Forward Equalizer (ANN-NFFE) and Deep Neural Network-Nonlinear Equalizer (DNN-NLE). Both the equalizers can mitigate the nonlinearities in the signal. Also, the impact of the system in downlink is analyzed using simulation by varying the data rates for various fiber lengths. The FBMC using SMT provides better spectral efficiency. The performance of both equalizers is compared. The addition of the DNN-NLE equalizer provides a better OSNR, good accuracy, and better BER compared with ANN-NFFE.