Availability-constrained shared backup path protection for GMPLS-based spare capacity reconfiguration

Abstract

Shared Backup Path Protection (SBPP) has been widely studied in the Generalized Multi-Protocol Label Switching (GMPLS) networks due to its efficient spare capacity sharing and flexibility in service provisioning. This thesis presents two policy-based models for evaluating the end-to-end (E2E) availability of an SBPP connection by assuming that no more than two simultaneous failures could possibly occur in the network. To minimize the redundancy while meeting the E2E availability requirement, a new parameter is defined for each connection, called protection level, which creates a framework of partial restoration from any unexpected failure. Based on the proposed availability model, two novel policy-based Linear Programming (LP) formulations are introduced - called failure-dependent and failure-independent policies, which aim to reconfigure the spare capacity allocation for dynamic provisioning of SBPP connections. Extensive simulations are conducted to validate the proposed availability model and demonstrate the effectiveness of the spare capacity reconfiguration architecture. The proposed availability-aware spare capacity reconfiguration (SCR) approaches are then implemented on top of a well known survivable routing scheme - Successive Survivable Routing (SSR), where the spare capacity saving ratio is taken as the performance measure. We show that the proposed SCR framework is an effective approach for achieving the GMPLS-based recovery in packet-switched networks.