Minimum delay path and capacity assignment for restoration in mesh and ring-mesh optical networks

This dissertation deals with optimum protection and restoration of all optical mesh networks using dense wavelength division multiplexing. Presently many optical networks are ring based. However, more and more mesh networks are being implemented in backbone networks. These networks carry high data rates and a failure lasting even a short time may lead to loss of large amount of data. There are proposed or developed methods for design of survivable networks that minimize restoration capacity but not the restoration time. This dissertation formulates and solves restoration problems that minimize restoration time and/or restoration spare capacity. The restoration optimization process uses three objective functions to minimize Average Increased Delay due to Restoration (AIDR) and Sum of Traffic Weighted Delay for minimum restoration time design and restoration capacity for minimum spare capacity design. The constraints considered are limits on working capacity, working capacity and spare capacity and working capacity plus spare capacity for restoration. In each case the results obtained are the optimum restoration traffic on each restoration path, cost of spare capacity and AIDR and the results are compared among different designs. This involves preprocessing steps that include finding restoration paths, numerical formulation of the objective functions and constraints and inputting these to the linear/nonlinear optimization software Lingo 9.0. In order to obtain results that can be reliable the previous design process has been applied and results have been obtained for five different networks. Results have also been obtained to illustrate the dependence of AIDR on the amount of available spare capacity. The above mentioned results have been presented in tabular form for easy comparison. Obtained results show that the methods developed in this dissertation require smaller restoration times than other available methods. The restoration capacities needed in the developed methods are in most cases comparable to those needed in other available methods. Thus the methods developed here will lead to building of survivable high-speed networks with minimum restoration time and loss of data.