کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Simultaneous OR and XOR All-Optical Logic Gates in Photonic Crystals Using Artificial Neural Network

نویسندگان :

Ehsan Adibnia¹ Niloufar Hosseini² Tofiq Nurmohammadi³

1- University of Sistan and Baluchestan 2- University of Sistan and Baluchestan 3- Urmia University of Technology

کلمات کلیدی :

Photonic crystals،all-optical logic gate،machine learning،neural networks،FDTD

چکیده :

All-optical logic gates are fundamental building blocks for high-speed photonic integrated circuits. This letter introduces a neural-network-assisted design methodology to simultaneously realize OR and XOR logic operations within a single, compact photonic-crystal structure. A dataset is generated via finite-difference time-domain (FDTD) simulations by sweeping four key parameters: the background rod radius, the defect rod radius, and the phase of each coherent optical input. A fully connected feedforward neural network is trained as a surrogate forward model to predict the normalized output power at two designated ports. The trained model demonstrates excellent agreement with FDTD results and enables rapid exploration of the parameter space to maximize logic-level contrast. An optimized design is identified, achieving XOR functionality at one output port and OR functionality at the other. The realized logic contrast exceeds 30 dB for the XOR output and 15 dB for the OR output. Compared to conventional full-wave, brute-force optimization, the proposed workflow reduces the design time by approximately 95%. This approach is general and can be extended to the design of other multifunctional photonic devices and logic gates on various photonic-crystal platforms.