Approaches for survivable access and core networks: Dual-homing and clustering

We have seen recent growth of telecommunication infrastructures---both in wireless access and wired core networks. The issue that becomes critical is to provide fault-tolerant and continuous service across large customer bases. In this dissertation, we propose multiple fault-tolerance techniques. We specifically pursue for dual-homing and clustering (or partitioning) techniques in wireless access, sensor, and optical core networks.We first study a survivable network design problem using dual-homing for hierarchical wireless access network infrastructures. Given the available capacity, connectivity, and reliability at each level, the problem is to minimize overall connection cost for multiple requests such that the capacity, connectivity, and minimum survivability constraints are not violated. Our study is different than earlier research in regard to the coordination of multiple layers of access networks. The connectivity to the core network may be fully or partially dual-homed paths, or may be single-homed paths. We study an off-line genetic algorithm based meta-heuristic.More often than not, wireless sensor networks (WSNs) are deployed in adverse environments where failures of sensor nodes are regular phenomena. Therefore, the organization or clustering of WSNs needs to be survivable. On the other hand, energy-efficiency in WSNs remains the main concern. In this work, we associate survivability and energy-efficiency in WSNs by dual-homing nodes during clustering and show that such a proactive scheme can actually increase the lifetime.In optical networks, physically disjoint routes are sought for failure-protection purpose, where the working path and the backup path of a request traverse diverse shared risk link groups (SRLGs). Hence, clustering of nodes should be effective in yielding SRLG-diversity. We address the problem of determining the appropriate clustering of nodes for a large wavelength division multiplexed (WDM) network. We find that considering risk group sharing in network partitioning best handles both scalability and survivability.In this work, we first introduce the concept of k-SRLG-connected partitioning for optical networks, where k&ge1. A logical topology is said to be k-SRLG-connected if its nodes remain connected upon any k-1 SRLG failures. We particularly focus on k=2. Simulation results verify that the proposed 2-SRLG-connected partitioning scheme performs better than traditional 1-connected partitioning approaches with regard to survivability and system stability.