Photonic Crystal-Based Binary-to-Gray Converter: A Low-Power Solution for Optical Neural Network Accelerators

This paper introduces an all-optical two-bit binary-to-Gray (B2G) code converter utilizing 2D photonic crystal waveguides, targeting efficient integration in binary neural network (BNN) hardware accelerators. The design leverages interference effects in a compact $10 \mu \mathrm{m} \times 10 \mu \mathrm{m}$ footprint, achieving extinction ratios of 9.49 dB and 10.11 dB for output bits G0 and G1, respectively, with a delay of 0.52 ps. This low-power, reversible device supports Feynman logic and BNN applications, offering a promising solution for optical computing. Simulations using the finite-difference time-domain method validate its performance.