Reliable Network Architectures: A Study on Provisioning, Routing, and Algorithms

With the increasing number of Internet users, connected devices, and high data rate applications, the future Internet traffic is expected to continue to grow rapidly. Able to keep up with this exponential growth rate are fiber optics network technologies that leverage wavelength division multiplexing (WDM) to achieve multiple orthogonal channels in every fiber. Achievable transmission rates for each wavelength channel are also improving, reaching now 400Gbps and above commercially. While the demand for high network throughput is going to be met by the emerging and foreseeable optical transmission technologies, network vulnerability to outages remains a concern, especially accounting for the aggregate large data traffic volume that can be carried by a single optical fiber or switched by a single optical network node. Backbone optical networks in both metro and wide areas carry large volumes of traffic and even a single network element outage (e.g., a fiber cut, a site power outage, a port failure) can cause significant traffic disruption and loss, which in turn can adversely affect a large group of network users. Therefore, protection and restoration schemes such as pre-planned restoration, fast reroute (FRR), and SRLG (Shared-Risk Link Groups)-aware protection need to be implemented to achieve the much-needed network survivability level. Some of these mechanisms have been vii used extensively in the past, but they are not able to provide high degrees of reliability that can prevent the massive service outages that have been experienced in some commercial networks in recent years, when two or more unexpected and concurrent events caused severe disruptions in the network. In this dissertation, a number of advanced protection and restoration schemes, along with their related optimization algorithms, are discussed and investigated, focusing on three optical network architectures -- GMPLS (Generalized Multiprotocol Label Switching) enabled WDM network, multilayer optical network, and virtual optical network. As discussed in the dissertation these advanced schemes and related optimization algorithms achieve high network reachability and survivability even in the presence of multiple failure events affecting the network. By utilizing these schemes and algorithms, optical networks can be designed and managed to achieve high successful-restoration rates, fast restoration completion time, high network resource utilization, and tolerances to one or more outages.