Abstract
Over the past decade, network traffic levels experienced an explosive growth at about double amount in approximately every thirty months. The sources accounting for this growth come from numerous high-speed applications (e.g., video-on-demand, high-definition television) which involve the data transmission in multicast groups. To realize optical multicasting, optical routers should have light splitters to split light signals and wavelength converters to change the wavelengths wherever needed. However, the splitting reduces the energy of the output signal which in turn requires the costly power amplification or regeneration. Wavelength converters are also immature to be deployed widely in current optical technologies. Consequently, in all-optical networks, routers are often heterogeneous in their processing units, which challenges the routing. Therefore, it is crucial to design efficient multicast routing strategies at the backbone optical networks, in order to achieve cost-performance tradeoff solutions while satisfying the ever-increasing bandwidth demands and optical constraints.In this thesis, we investigate the all-optical multicast routing (AOMR) problems in heterogeneous optical networks. The heterogeneity mainly comes from the absence/presence of light splitters and wavelength converters and the uneven distribution of wavelengths in the network links. In general, AOMR problems are often NP-hard. The objective of the thesis is to analyze and formulate the problems, to search for the optimal solutions, and to propose efficient heuristics to solve the problems under different optical constraints. Both possible contexts, i.e., single-multicast request and multiple-multicast requests, are examined. All the reported results in the thesis are supported by extensive and careful simulations. The major contributions can be summarized as follows.1) We identify the optimal route structures for AOMR problems under heterogeneous mesh WDM networks. As shown in the thesis, the optimal solutions are no longer based on conventional light-trees, but a more general tree-like structure called hierarchy. Some forms of hierarchy realized for WDM multicasting are light-trails, light-hierarchies, light-spider hierarchies and a set of these light-structures. The exact and heuristic algorithms proposed in the thesis are mainly based on hierarchy. 2) For single-multicast with sparse-splitting case, we propose an efficient heuristic algorithm to produce a good tradeoff solution among wavelength consumption, channel total cost and end-to-end delay.3) For single-multicast with non-splitting case, we prove the NP-hardness, identify the optimal solution as a set of light-spider hierarchies, formulate the problems by means of Integer Linear Program (ILP) formulations to find the exact solution, and propose several cost-effective heuristic algorithms to compute the approximate solutions. 4) For the case with multiple-multicast requests, we focus on static traffic patterns under sparse-splitting without wavelength conversion case. First, an ILP formulation based on light-hierarchies is proposed to search for the optimal solution. By applying the layered graph model, we then develop several adaptive heuristic algorithms to compute light-hierarchies for approximate solutions. These adaptive algorithms outperform the existing fixed routing ones in minimizing the blocking probability. Overall, the thesis points out that the optimal solutions for heterogeneously constrained AOMR problems correspond to hierarchies, regardless of request multiplicity consideration.