Multi-layer Architecture and System Design of Internet Protocol (IP) and Optical Networks

Abstract

The modeling of multi-layer networks is well-researched problem, but existing models for IP-optical integration do not consider technology specific capabilities and operational aspects for employing dynamic optical circuits in IP networks. This thesis presents an ILP-based model which identifies and incorporates novel constraints for numerous technology specific aspects, such as IP forwarding capabilities and behavior of routing protocols. Novel solutions for critical operational aspects of IP-optical integration, such as optical circuit decommissioning and computation under unknown traffic conditions, are also proposed in this thesis. The thesis identifies changes in routing as a major deterrent for employing dynamic optical circuits in IP networks, and proposes the new Optical Bypass approach to address the same. Quantitative studies presented indicate that the introduction of an optical circuit under this approach significantly reduces the effect on IP routing, while lowering optical capacity requirements as compared to the traditional SPF based approaches. The proposed solution can also compute near-optimal solutions under unknown IP traffic matrix conditions, making it ideal for application in dynamic network scenarios. The thesis also addresses specific management challenges with IP-optical integration, and outlines solutions to address the same. The solutions are built around enabling coordination of management subsystems in the two network layers. The thesis presents the general architecture to facilitate coordination between management subsystems in a programmable fashion and demonstrates the capability of the architecture to be used in legacy as well as SDN-capable infrastructure. The thesis also outlines the design and implementation of the first open-source PCE, which is a critical management subsystem for enabling multi-layer path computation in IP-optical networks.