Machine learning-based inverse design of raised cosine few mode fiber for low coupling

In mode division multiplexing techniques, few-mode fiber (FMF) with multiple modes and low coupling is the demanded design. The design of FMF become more complex due to different parameters and shape design of it profile especially when you need to get a design with high number of modes, and it will...

Full description

Saved in:
Bibliographic Details
Published in:Optical and quantum electronics 2024, Vol.56 (1), Article 56
Main Authors: Chebaane, Saleh, Ben Khalifa, Sana, Jebali, Maher, Louati, Ali, Bahri, Haythem, Dafhalla, Alaa
Format: Article
Language:English
Subjects:
Characterization and Evaluation of Materials
Computer Communication Networks
Computing time
Coupling
Design parameters
Electrical Engineering
Fiber optics
Inverse design
Lasers
Machine learning
Multiplexing
Neural networks
Optical Devices
Optics
Photonics
Physics
Physics and Astronomy
Propagation modes
Refractivity
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In mode division multiplexing techniques, few-mode fiber (FMF) with multiple modes and low coupling is the demanded design. The design of FMF become more complex due to different parameters and shape design of it profile especially when you need to get a design with high number of modes, and it will consume a lot of computing time. Therefore this paper propose a machine learning technique employ neural network to design FMF raised cosine profile inversely. We realize the inverted design of raised cosine with U refractive index layer FMFs for supporting 6 mode operation, by adopting the minimal index difference between neighboring modes to get low coupling between modes. This proposed strategy gives higher precision with less complex design of FMF, in addition to low coupling between propagating modes. This promoting method can be applied in other types of fiber optic profile which it need a lot of parameters optimization.
ISSN:0306-8919
1572-817X
DOI:10.1007/s11082-023-05695-8