Aprimoramento de modelos para a otimização de amplificadores Raman

Abstract

The emergence of optical amplifiers in the nineties, enabled the use of wavelength multiplexing in optical communication systems and the increase in the transmission rates. The Raman amplifiers is inserted in this category of optical amplifiers. This technology allows the amplification of several wavelength simultaneously and transparently. However, it presents the difficulty of obtaining a plain amplification band, specially when this bands started to increase in extension. It was based in this problem that this dissertation proposes a optimization method based in a hybrid algorithm, that mixes components of genetic algorithms with a geometric compensation technic, in order to optimize the parameters of Raman amplifiers, making it possible to achieve certain project requirements. This means one should determine the smallest number of laser pumps necessary to meet the project parameters of minimal gain required, and the maximum ripple allowed by the system, inside a wide signal bandwidth, like the C + L band. This study verified that the hybrid algorithm is capable of finding efficient solutions of plain bandwidth, even in wide band conditions (C + L band), obtaining better results than those obtained with pure genetic algorithms applied to the same problem.