MACHINE LEARNING APPLICATION IN INVERSE DESIGN OF FEW-MODE FIBERS

Abstract

The importance of optical fiber research is increasing due to its applications in the digital world, including components, sensors, and high data rate communication. Few-mode fiber (FMF) research is regenerating due to its high data rate transmission ability. This dissertation work proposes new designs of FMFs with updated material composition and geometry to establish weakly coupled spatial division multiplexing (SDM)/mode division multiplexing (MDM) links. The next generation of communication, 5G aims to connect people and things via intelligent networks, but current network architectures struggle to handle massive data traffic. The spatial domain of the fiber is highly useful for handling this massive data traffic. This work reviews the requirements of 5G networks and how they can be handled through spatial multiplexing and mode multiplexing through a few-mode optical fiber. The article demonstrates machine learning-based inverse modeling of the triangular-ring-core few-mode fiber profile with weak coupling optimization.