Abstract

Fiber optic technologies enabling high-speed, high-capacity digital information transport have only been around for about 3 decades but in their short life have completely revolutionized global communications. To keep pace with the growing demand for digital communications and entertainment, fiber optic networks and technologies continue to grow and mature. As new applications in telecommunications, computer networking and entertainment emerge, reliability, scalability, and high Quality of Service (QoS) requirements are increasing the complexity of optical transport networks. This dissertation is devoted to providing a discussion of existing and emerging technologies in modern optical communications networks. To this end, we first outline traditional telecommunication and data networks that enable high speed, long distance information transport. We examine various network architectures including mesh, ring and bus topologies of modern Local, Metropolitan and Wide area networks. We present some of the most successful technologies used in todays communications networks, outline their shortcomings and introduce promising new technologies to meet the demands of future transport networks. The capacity of a single wavelength optical signal is 10 Gbps today and is likely to increase to over 100 Gbps as demonstrated in laboratory settings. In addition, Wavelength Division Multiplexing (WDM) techniques, able to support over 160 wavelengths on a single optical fiber, have effectively increased the capacity of a single optical fiber to well over 1 Tbps. However, user requirements are often of a sub-wavelength order. This mis-match between individual user requirements and single wavelength offerings necessitates bandwidth sharing mechanisms to efficiently multiplex multiple low rate streams on to high rate wavelength channels, called traffic grooming. This dissertation examines traffic grooming in the context of circuit, packet, burst and trail switching paradigms. Of primary interest are the Media Access Control (MAC) protocols used to provide QoS and fairness in optical networks. We present a comprehensive discussion of the most recognized fairness models and MACs for ring and bus networks which lay the groundwork for the development of the Robust, Dynamic and Fair Network (RDFN) protocol for ring networks. The RDFN protocol is a novel solution to fairly share ring bandwidth for bursty asynchronous data traffic while providing bandwidth and delay guarantees for synchronous voice traffic. We explain the light-trail (LT) architecture and technology introduced in [37] as a solution to providing high network resource utilization, seamless scalability and network transparency for metropolitan area networks. The goal of light-trails is to eliminate Optical Electronic Optical (O-E-O) conversion, minimize active switching, maximize wavelength utilization, and offer protocol and bit-rate transparency to address the growing demands placed on WDM networks. Light-trail technology is a physical layer architecture that combines commercially available optical components to allow multiple nodes along a lightpath to participate in time multiplexed communication without the need for burst or packet level switch reconfiguration. We present three medium access control protocols for light-trails that provide collision protection but do not consider fair network access. As an improvement to these light-trail MAC protocols we introduce the Token LT and light-trail Fair Access (LT-FA) MAC protocols and evaluate their performance. We illustrate how fairness is achieved and access delay guarantees are made to satisfy the bandwidth budget fairness model. The goal of light-trails and our access control solution is to combine commercially available components with emerging network technologies to provide a transparent, reliable and highly scalable communication network. The second area of discussion in this dissertation deals with the rapid prototyping platform. We discuss how the reconfigurable rapid prototyping platform (RRPP) is being utilized to bridge the gap between academic research, education and industry. We provide details of the Real-time Radon transform and the Griffin parallel computing platform implemented using the RRPP. We discuss how the RRPP provides additional visibility to academic research initiatives and facilitates understanding of system level designs. As a proof of concept, we introduce the light-trail testbed developed at the High Speed Systems Engineering lab. We discuss how a light-trail test bed has been developed using the RRPP to provide additional insight on the real-world limitations of light-trail technology. We provide details on its operation and discuss the steps required to and decisions made to realize test-bed operation. Two applications are presented to illustrate the use of the LT-FA MAC in the test-bed and demonstrate streaming media over light-trails. As a whole, this dissertation aims to provide a comprehensive discussion of current and future technologies and trends for optical communication networks. In addition, we provide media access control solutions for ring and bus networks to address fair resource sharing and access delay guarantees. The light-trail testbed demonstrates proof of concept and outlines system level design challenges for future optical networks.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call