Abstract

Survivability of high-capacity optical wavelength-division multiplexing (WDM) mesh networks has received much research attention for many years now. These networks are typically designed to survive single component failures. The method of pre-configured protection cycles ( p -cycles), recently proposed by W. Grover's research group, promises to achieve ring-like high speed protection with mesh-like high efficiency in use of spare capacity. In such networks, which are designed to withstand only single failures, service availability comes to depend on dual-failure (or more) considerations. Hence, availability-aware service provisioning emerged as a topic of great importance in the past few years. In this thesis, we first revisit the problem of availability analysis in p -cycle based networks and present an accurate model for availability-aware provisioning after highlighting major flaws in prior work. Our model provides a technique for allocating p -cycles to restore single link failures such that the unavailability of all the demands in the network is bounded by an upper limit. We then provide some heuristics for restricting the number of variables and constraints in an integer linear programming formulation in order to solve our problem in a reasonable amount of time. Failure-Independent Path Protecting (FIPP) p -cycle recently has been proposed as an extension of the basic p -cycle to provide a pre-connected, failure independent, path-protecting network design. We present in this thesis the first model for availability-aware provisioning in FIPP based networks. Our study focuses on determining whether FIPP will maintain its resource efficiency advantages over span p -cycles when the network design is based on limiting the service unavailability.