Efficient Design and Management of Reliable Optical Networks

Optical networks have played a major role in allowing us to meet the bandwidth demands driven by the explosive growth of the Internet. As more and more bandwidth-hungry applications emerge, the aggregated traffic at the backbone layer will continue to grow. This increase in capacity poses challenges, especially in terms of reliability and overall management overhead. At the same time, the high bandwidth available in backbone networks lead to an increase in vulnerability to failures as huge amounts of data can be lost in a relatively short period of time. In order to meet the bandwidth demand as well as the performance requirements of future network services, cost-effective solutions that guarantee a desired level of reliability at the core must be provided. To this end, understanding the limitations of various survivability algorithms and protocols is important. Given the cost of physically modifying the backbone, as well as the relatively slow speed at which core architectures change compared to higher layers, it is important to be able to optimize the network using soft solutions to achieve better performance. A few important aspects of network reliability and failure management are studied in this thesis. First, a study on the impact of multiple failures and techniques for addressing such failures are presented. Second, service differentiation based on reliability needs is studied. Service differentiation is critical in balancing network operation costs as well as in maximizing network utilization. Third, a cost-efficient, high-speed recovery scheme is introduced and evaluated under online non-dynamic and dynamic scenarios, as well as static provisioning applicable to protecting mission-critical traffic. Finally, the impact of physical layer impairments on network reliability is presented