Obtenção de portas lógicas totalmente ópticas usando redes neurais

Abstract

This work presents a study on the use of neural networks MLPs (from English Multilayer Perceptrons) to study the relationship between the physical parameters and the logical characteristics of a PCF. The modulation used was PAM (from English Pulse Amplitude Modulation) and OOK (from English On-Off Keying). In this problem, the variations of specific combinations of the values of physical parameters determine the logic pulses of outputs of this PCF. Such combinations were analyzed by a neural network MLPs, and this was able to accurately classify the physical characteristics that determine the occurrence of the OR logic gate. The neural network is able to identify specific logic gates based only on the physical parameters that generate these outputs. The obtained results show that the neural network is able to learn the periodic behavior and the existing relationships between the physical parameters: coupling constant k, propagation length z and amplitude modulation parameter m. Of the Boolean expressions analyzed, the OR logic gate stands out, whose signals were classified with a precision of 98,71%. In this way, the analysis and the results described on this coupling device demonstrate that the proposed RNA is capable of recognizing relationships between the physical parameters and the output signals of a PCF logical device and that MLPs are tools of great importance and efficiency for the recognition of these patterns, helping telecommunication transmission systems.