Towards Improving Resilience of Network Slices Through Protection and Localization

The scope of the telecommunication industry is expanding due to the inclusion of a diverse range of applications that requires a dynamic resource provisioning mechanism to accommodate the changes over time while also ensuring robustness against failures. These applications are deployed on the shared physical infrastructure through network slicing, which partitions the available shared resources, customizing them for specific services. Network slicing is supported through enabling technologies like Software Defined Networking (SDN) and Network Function Virtualization (VNF). SDN aids the network operators with the ability to manage the traffic effectively, whereas, NFV helps in the cost-efficient migration of network functions. The ability of network slices to continue to support the service requirements of applications is enabled by methodologies that improve the robustness by utilizing cost-effective measures. This dissertation explores four problems towards robust network design from the perspective of network operators. Conventional approaches to ensuring robustness in networks entail protection and recovery mechanisms against single network component failures. However, addressing robustness against multiple network component failures introduce additional challenges from the perspective of the network operator. To address this issue, the first problem discusses a protection mechanism to deploy survivable routing in multi-domain networks against geographically correlated failures. The second problem discussed here is a localizavii tion technique to deduce which set of physical links have failed by monitoring the state of the network slices (virtual links) deployed on the physical infrastructure. Finally, the next two problems ensure that the availability of the services is satisfied. In particular, a network slice composition problem is presented for specific applications while guaranteeing the availability requirements of the services and minimizing the cost of resources used. The ideas discussed in this dissertation utilize simple yet efficient algorithms which gear the network operators with cost-efficient techniques to provide resilience to its users.