Optimization of gain flattened fiber Raman amplifier model with binary search equation based adaptive artificial bee colony (BSEAABC) algorithm


Yolcu V., Yücel M., Aydın D.

JOURNAL OF THE FACULTY OF ENGINEERING AND ARCHITECTURE OF GAZI UNIVERSITY, vol.39, no.1, pp.29-38, 2024 (SCI-Expanded) identifier identifier

Abstract

In this study, a distributed fiber Raman amplifier (FRA) system was established by sending 4, 8, and 16 pump signals respectively in the opposite direction to 100 optical signals advancing on a fiber line with a distance of 50 km (Table A). In order to flatten the signal gain in the established system, the non-linear FRA model was defined as an optimization problem and solved with the binary search equation based adaptive artificial bee colony algorithm (BSEAABC). The optimum pump powers and wavelength values found are presented below.Purpose: In this study, the optimum wavelengths and frequencies of the pump signals sent to the system for 100 optical signals, whose gain was tried to be flattened in SMF-28 type fibers, were found with the binary search equation based adaptive artificial bee colony algorithm (BSEAABC).Theory and Methods: The most important interactions in the simplified model of FRAs; pump-pump, pump-signal interactions, together with the wavelength-dependent attenuation coefficient acting on the signals and the pumps. In the study, the nonlinear differential equation of FRA was solved. In the established FRA system, classical Raman gain coefficient and attenuation coefficient of SMF-28 optical fiber were used.Results: In this study, the optimum wavelengths and frequencies of the pump signals sent to the system for 100 signals whose gain was tried to be flattened in SMF-28 type fibers were solved with the BSEAABC algorithm. When the results were examined, the average net gain difference obtained by using 4 pumps was found to be 0.16 dB, and the net gain difference was found to be 0 & PLUSMN; 0.4dB. The average net gain difference obtained by using 8 pumps was 0.043 dB, and the net gain difference was 0 & PLUSMN; 0.1dB. The average net gain difference obtained by using 16 pumps was determined as 0.028 dB, and the net gain difference was determined as 0 & PLUSMN; 0.05 dB. Conclusion: It is thought that in parallel with the development of algorithms produced within the scope of artificial intelligence technology, solution performances similar to this study will improve and thus the efficiency obtained will increase.