Design of a compact all-optical digital-to-analog converter based on photonic crystals using neural networks

Abstract

This paper presents the design and optimization of a compact 2-bit all-optical Digital-to-Analog Converter (DAC) based on photonic crystals (PCs), utilizing an Artificial Neural Network (ANN) for enhanced performance. The proposed structure features a square lattice photonic crystal composed of silicon rods in an air background, designed to operate within the photonic bandgap for the TM mode at a wavelength of 1.55 μm. The dimensions of defect rods in the photonic crystal structure are optimized using a feed forward ANN, trained with simulated data to maximize the output power of the DAC. The ANN model demonstrates high precision in predicting the optimal dimensions, achieving a significant reduction in design size and complexity. The designed DAC performance is validated through simulations using two software, and the results are consistent with predictions, confirming the validity of the design. The proposed 2-bit DAC exhibits superior performance with minimal footprint of 116 µm², making it highly suitable for integration into all-optical digital circuits and next-generation optical fiber technologies.