Optical hybrid switching with flow classification in IP over optical network

Abstract

In this dissertation, we studied the optical hybrid switching which is a new switching technique for the future optical Internet. The first part of the dissertation, we consider a specific IP-centric optical IP network architecture which has a hierarchical Quality of Service (QoS) mapping model. We propose a new optical hybrid switching technique which combined Optical Burst Switching (OBS) and Optical Circuit Switching (OCS) using flow-level service classification scheme. In particular, this classification scheme classifies incoming IP traffic flows into short-lived and long-lived flows for QoS provisioning according to traffic characteristics. In the proposed hybrid switching, the short-lived flows including delay sensitive traffic use OBS and the long-lived flows including loss-sensitive traffic use OCS. Therefore, optical hybrid switching takes advantages of both switching technologies. We propose the detailed flow classification algorithm and QoS provisioning mechanism for optical hybrid switching. The aim is to maximize network utilization while satisfying user’s QoS requirements. The second part of the dissertation, we analyzed the performance of OBS and optical hybrid switching. For OBS, we demonstrate that if no wavelength conversion is available loss-based OBS systems are not efficient. Although some efficiency improvement is achieved by burst segmentation, without significant wavelength conversion, high efficiency is not achievable. While the burst size and the assembly time have a potent influence on the delay performance, OBS has more advantages for delay than OCS due to one-way reservation. Through the end-to-end delay analysis, we give you a good guideline to find the optimal burst size and offset time for efficiently operating optical hybrid switching. For optical hybrid switching, we model the system as a single server queue in a Markovian environment. The burst generation process is assumed to follow a two-state Markov Modulated Poisson Pr...