Cost -efficient design of waveband switching in optical wavelength division multiplexed networks

As the demand for bandwidth grows drastically on the Internet, it drives the evolution of current networks, especially the current long-haul networks. With the development and deployment of advanced network technologies, low-cost and high-performance next-generation networks will be practical in a few years. Among all of the advanced technologies for wide-area backbone networks, wavelength-division multiplexing (WDM) is the most promising candidate recognized both by the academic and industrial communities. Through WDM, data is transmitted as optical signals, which are switched from input ports to output ports based on the wavelengths at intermediate nodes. To save the utilized switching ports and to reduce the nodal complexity and size, waveband switching (WBS) technology is introduced in WDM networks to group and switch a specific set of wavelengths as a single bundle. This dissertation is focused on the design and operation of a WDM network, which supports waveband switching of data. Our research shows the advantages of adopting the WBS technique in saving the operational expenditures and in reducing the complexities of optical components for backbone networks. In this dissertation, we first investigate the impacts of optical component developments on waveband switching in WDM networks. By adding banding filters into current optical add-drop multiplexers, wavebands can be multiplexed with wavelengths into a fiber. Through embedding an all-optical waveband switching fabric into an optical crossconnect (OXC), data in waveband signals could be switched and routed. Second, to assist and simplify the design of cost-efficient waveband switching algorithms, a generic auxiliary graph model is proposed, which considers the constraints on waveband switching. The changing network configurations are captured by the auxiliary graph, which can provide detailed information to efficiently provision connections. Third, waveband switching approaches with different grouping strategies, namely end-to-end grouping and intermediate grouping, are proposed. Heuristic algorithms are developed accordingly to take full advantage of the WBS functions by adopting a load balancing technique. Fourth, stochastic mathematical models are formulated to analyze the proposed waveband switching algorithms in terms of network throughput. The analytical results show good matches with the simulation results. The analyses also reveal some important insights into how to achieve optimal performance under the current network configurations. Fifth, to ensure the survivability of the network provisioning under a single network component failure, two dedicated-path protection schemes are proposed, which consider the shared-risk link group constraints. To maximize network throughput and minimize network provisioning costs, integer linear programming (ILP) formulations are developed and heuristics adopting random optimization are developed. Finally, a more practical heterogeneous waveband switching approach is proposed, which clusters the network into multiple autonomous systems (ASs). The AS may contain some specific nodes that provide waveband grouping and waveband disaggregating functions