Optical Network Resource Allocation Algorithm

All-optical network can alleviate the O/E/O conversion bottleneck by switching and routing in optical field. The resource of assignment problem includes optical hardware, routing and wavelength. With static demand in all-optical network, the static routing and wavelength assignment problem studies the way of routing and wavelength assignment to minimize the resource consumption.In all-optical network the destruction of single link or node will lead to communication losing, and protecting strategy is used to keep the communication fluently. The ILP model of path-level protection is proposed to protect the static routing and wavelength. Furthermore traffic grooming reduces the resource consumption in all-optical network by grooming the low rate demands in high transport channel. The grooming way will influence the results of routing and wavelength assignment, thus the traffic grooming model is constructed to solve the assignment problem.The means of solve the static routing and wavelength assignment problem includes construction the ILP model and heuristic algorithm. The solution of heuristic algorithm is close to the optimal assignment and time to solve is quickly. But for large-scale network, the performance of heuristic algorithm can not been evaluated.In chapter 2, the paper constructs the ILP model for protection and traffic grooming to solve the static routing and wavelength assignment problem. For large-scale network, Lagrange relaxation algorithms, Dual and Primal decompositions are proposed. The test by computer software shows that ILP model can get the optimal assignment. But the problem is NP-C problem, for large-scale optimization the time to solve the model is too long and even can not get the feasible solution. Decomposition theory provides the mathematical way to build an analytic foundation for the design of modularized and distributed control of networks. On theory the decomposition have the same solution as ILP model. The test validates that after limited iterations, the solutions of decomposition are near-optimal and the time used solving the model is more quickly than the ILP model.Multicast in all-optical replicates and delivers the optical signal by the splitter to reduce the resource used in the optical network. Placing splitters at all of the nodes is considered to be expensive in the all-optical WDM network. With fixed number of splitter, optimally selecting a subset of nodes to place the splitters can reduce the cost and this placement problem is called the splitter placement problem. In Chapter 3, with the static demand in network, ILP model and MF algorithm are used to solve the splitter placement problem. The test in the paper validates that the ILP model can gain the optimal splitter placement allocation, but the time used to solve the model is too long and can't get the feasible allocation for large-scale network. MF algorithm can get the splitter placement allocation close to optimal allocation more quickly and does not depend on the algorithm of construct the light-tree.In Chapter 4, with forecast uncertain demands and a fixed number of splitter, a robust optimization strategy is proposed to optimally solve the splitter placement problem for multicast in all-optical networks. The test in the paper shows that using robust strategy, the resource provisioning of each scenario is closer to the optimal resource consumption, more robust, less resource consumption and more practically than the static solution. Chapter 5 concludes the whole paper and future work.