Resource allocation in wavelength -routed WDM mesh networks

Wavelength-routed WDM networks have been considered as the best candidate for the next-generation telecom backbones. In a wavelength-routed network (WRN), fiber links are interconnected by a set, of optical cross-connects (OXCs). Using the wavelength-division-multiplexing (WDM) technology, a fiber link can simultaneously carry multiple wavelengths, each of which provides bandwidth up to 40 gigabits per second (Gbps). Such huge bandwidth eliminates concerns about network capacity imposed by the explosive expansion of Internet traffic. However, how to efficiently utilize network resources remains a challenge. This dissertation addresses the resource allocation problem in wavelength-routed WDM mesh networks. The goal of this dissertation is to model the wavelength-routed networks which form the next-generation backbone infrastructure, formulate resource allocation problems in such networks, and develop cost-efficient solutions to solve these problems. In particular, this dissertation addresses three topics. The first topic is that of shared risk link group (SRLG)-diverse path provisioning under hybrid service level agreements in wavelength-routed mesh networks. In this topic, the static provisioning problem is incorporated with some realistic constraints, including multiple classes of protection schemes under the SRLG-diversity constraint, the lightpath length constraint, and the revenue value of each connection. The problem is formulated as two sub-problems revenue maximization and capacity minimization problems. Integer linear programming (ILP) formulations are developed to model these problems. A rerouting and a tabu search heuristic are developed to solve these problems. The experimental results show that the tabu search heuristic outperforms the rerouting approach. In the second topic, a novel load-balancing protection path re-optimization approach is developed to postpone network upgrade, resulting in significant cost savings. In a network with dynamic traffic, resource utilization becomes sub-optimal as the network evolves with arrival and removal of connections. Re-optimization, which is referred to as the method of re-arranging network resources, is an attractive solution for this issue. The dissertation defines a novel concept, load distribution vector (LDV), and sets the re-optimization objective as minimizing the network LDV. A link-based greedy heuristic is developed to achieve this goal. The experimental results show that the proposed approach achieves significant cost savings over a previously proposed approach in both uniform and non-uniform networks. (Abstract shortened by UMI.