| The explosive growth of Internet traffic and emerging applications, such as multimedia delivery and Grid computing, is driving tremendous demands for bandwidth capacity and reliable transfer in long-haul transport networks. In this work, we focus our research on survivability mechanisms for the multiple-failure scenario and emerging Grid applications. The major challenge under multiple failures is to deal with failure combinations in exponential complexity. In this work, we address several problems of maximizing end-to-end reliability, called MAX-REL, under multiple failures in general mesh networks.;First, we investigate the issues of how reliable two-path protection can be and how to achieve the maximum reliability. Solving MAX-REL provides 100% protection against a single failure while maximizing the reliability regardless of how many link failures occur in the network. We prove that this problem is NP-Complete and derive a corresponding upper bound and a lower bound. Simple performance-effective and time-efficient heuristic algorithms are proposed.;Next, we consider the subject of reliability using segment protection, which is expected to tolerate more failures and achieve higher reliability than path protection. Two categories of segment protection, non-overlapping segment protection and overlapping segment protection are investigated.;We then extend independent failures to Shared Risk Link Groups, which introduce failure dependency on different logical links in layered networks. In this scenario, multiple generic risks are assumed, where each risk is a generic failure event, which may result in single-link failure, multiple-link failure, or node failure. A novel graph model and corresponding heuristic algorithm are proposed.;A major challenge for Grid Computing is how to locate computing resources distributed within the networks. Correspondingly, network survivability mechanisms for Grid Computing provide survivable and reliable transmission to indefinite destinations. We address the issues of survivability within two specific architectures.;We investigate survivable traffic grooming for anycast over Wavelength Division Multiplexing (WDM) networks. An auxiliary graph model and three differentiated survivable schemes based on the graph model are proposed.;At last, we focus on reliable multi-resource manycast over Optical Burst Switching (OBS) networks. Reliable transmission mechanisms are proposed to minimize the resource blocking probability, which results from burst contention in the OBS network and from resource unavailability at the destinations. An analytical model for various schemes is developed. |
