Improved and Efficient Optical Networks On Chip Architectures and Designs

Abstract

Optical Network-on-chip (Optical-NoC) is a promising and emerging on-chip optical interconnect technology to develop energy-efficient and high-performance inter-core communication for many-core processors. By utilizing the silicon interconnects to transmit optical signals, it can achieve ultra-low communicational latency, high bandwidth capacity, and low energy consumption. While several studies have been conducted on designing optical-NoC architecture and routing schemes to support proper communication, most of the existing solutions adopt the methods that were used for electrical interconnects. These solutions limit to achieve the full advantage of optical-NoC communications. Moreover, most of them focus only on their proposed architecture and routing schemes used to perform communication on those architectures. Hence, this thesis shall address the design of an efficient communication scheme for optical-NoC architectures by taking into account the distinctive characteristics and constraints of an optical-NoC. newlineThis thesis examines and studies the problem from a network-level perspective. The design methodology is to deploy optimal routes between PE nodes, with the objective of efficiently utilizing optical waveguide deployment. The novelty is to adopt an optical NoC architecture, which helps to act as a brick-like structure and helps to configure other big optical-NoC architectures. As per routing schemes, two routing algorithms are proposed to fulfill different communication requirements. newlineTo begin with, the photonic integrated circuits (PICs) or optical networks on chip (Optical-NoC) are proposed, which helps to provide a promising solution for on-chip optical communication. Optical-NoCs are providing a better solution with respect to the traditional NoC interconnects which face major challenges such as latency, bandwidth, and energy consumption. Optical-NoCs have the ability to provide low power communication and high data processing at ultra-high speeds.