Open AccessFuzzy-Neuro Network in a CO-OFDM system: Various Membership Functions Comparison
Author(s) -
Gurpreet Kaur,
Gurmeet Kaur
Publication year - 2021
Publication title -
aijr proceedings
Language(s) - English
Resource type - Conference proceedings
ISSN - 2582-3922
DOI - 10.21467/proceedings.114.46
Subject(s) - orthogonal frequency division multiplexing , sigmoid function , gaussian , membership function , backpropagation , nonlinear system , activation function , fuzzy logic , gaussian function , function (biology) , division (mathematics) , mathematics , artificial neural network , fuzzy set , computer science , algorithm , artificial intelligence , statistics , arithmetic , physics , evolutionary biology , biology , estimator , quantum mechanics
Fuzzy-Neuro Network based nonlinear equalizer (FNN-NLE) has been used for the extenuation of nonlinearities in optical communication systems. Until now, many membership functions with resilient backpropagation activation function was used for making FNN-NLE in a coherent optical orthogonal frequency division multiplexing (CO-OFDM) systems. Despite this, no research is reflecting the comparison of different membership functions (MFs). In this paper, various membership functions such as gaussian MF, gaussian combination MF, triangular MF, difference between two sigmoidal functions MF, pi shaped MF, generalized bell shaped MF, trapezoidal MF and product of two sigmoid functions MF has been compared. From this study, the maximum performance in terms of BER is achieved with gaussian membership function has been concluded.