Projeto ótimo de moduladores eletroópticos / Optimum design applied to electrooptic modulators

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

Many design developments involve the optimization of several objective functions. These objective functions are often conflicting with each other and they cannot be simply aggregated into only one function to be optimized. This type of problem is known as the multiobjective optimization problem, for which the solution is a set of points known as Pareto optimal set. The multiobjective problems are often difficult to solve without the aid of efficient optimization algorithms. This work presents a new approach for multiobjective optimization, using a genetic algorithm, applied to the electrooptic modulators optimization. These modulators are key devices for external modulation in fiber optics communication system. The device analyzed in this work is a Mach‑Zehnder modulator made of titanium diffused lithium niobate. To solve the direct problem, the finite elements method is applied, which is adequate to cope with complex geometries and non homogeneous and anisotropic materials. To solve the inverse problem, two multiobjective techniques were implemented, the weighted sum method and the e-constraint method, to optimize the electrical characteristics of the modulator in order to achieve an optimal geometry. The optimization of pairs of characteristics allowed obtaining geometrical configurations that present large bandwidth e low drive power. Additionally, it was shown that some pairs of characteristics generate complex Pareto frontiers with nonconvex regions. The optimization process implemented is efficient and robust, and the analysis mechanism proposed allows a considerable reduction of cost and time of development.

ASSUNTO(S)

finite elements method algoritmos genéticos genetic algorithm computação aplicada moduladores eletroópticos otimização multiobjetivo multobjective optimization método dos elementos finitos electrooptic modulators computer science

Documentos Relacionados