Resource allocation and management is a key problem in optical networks. It covers many different topics such as routing and wavelength assignment (RWA), traffic grooming and load balancing, as well as provisioning and protection. Numerous algorithms have been proposed to address those issues. But these algorithms treat each topic separately and fail to realize the similarity and consistency of the ONRMA (optical network resource management and allocation) problem, which in turn hinders the practical implementation of those schemes....[ Read more ]

Resource allocation and management is a key problem in optical networks. It covers many different topics such as routing and wavelength assignment (RWA), traffic grooming and load balancing, as well as provisioning and protection. Numerous algorithms have been proposed to address those issues. But these algorithms treat each topic separately and fail to realize the similarity and consistency of the ONRMA (optical network resource management and allocation) problem, which in turn hinders the practical implementation of those schemes.

In general, there are several issues that must be dealt with in the ONRMA problem. Usually there are many sources distributed in the network competing for the use of the available bandwidth, we need to address: 1) the efficient bandwidth allocation to the different requests taking into account their different needs and performance requirements; 2) the fairness among different nodes; 3) the implementation of the allocation scheme with minimal communication overhead; and the 4) QoS requirements and bandwidth pricing according to the service providers.

In this thesis, we propose a general and powerful framework to address the ONRMA problem. It can efficiently plan the allocation of network resources to connection requests and provide a novel way of identifying the bottlenecks in the network. In particular, by dividing the network into different clusters (islands), we build a network hierarchy which reduces the complexity of the problem and make it possible to formulate several heuristics to guide the search towards a solution. This framework is based on a clustering scheme called Blocking Island.

In this thesis, we firstly introduce some background of optical network resource management and allocation problem. We then study the lightpath establishment in WDM optical networks, which is also known as the routing and wavelength assignment (RWA) problem. Inspired by the idea of Blocking Island, we design a network model called Blocking Island Graph (BIG), which can be employed to solve many issues of optical network resource management and allocation. Based on this model, a generic RWA algorithm, BI_RWA, is proposed. This algorithm can solve the RWA problem under different assumptions: static or dynamic traffic, and single or multiple fiber links between node pairs. We then extend the BI_RWA algorithm to accommodate Multi-segment optical networks. We also demonstrate that the BIG network model can be used to optimize the placement of limited number of wavelength converters in optical networks. With the development of Internet, the focus of traffic grooming is evolving from ring topology to mesh topology. We present an ILP (integer linear programming) formulation to minimize the network cost (transceivers and wavelength). We also propose a new heuristic based on the Blocking Island paradigm for traffic grooming in mesh networks. In terms of optical network survivability, we put our emphasis on IP over WDM networks. We propose an integrated IP over WDM routing and grooming algorithm in GMPLS based optical networks and we also design a dynamic provisioning/protection scheme in IP over WDM networks.