This paper investigates design methods of protection schemes in survivable WDM networks which use pre-configured protection structures (p-structures) in order to provide different quality-of-recovery (QoR) classes within 100% resilient single link protection schemes.

The design problem of a protection scheme in a survivable WDM network corresponds to a trade-off among network cost, recovery delay, and management complexity. In order to strike a balance between those design parameters, differentiation of the quality-of-recovery is the practical and effective approach to lower the protection cost while providing the different services. Prevalent design approaches of protection schemes are based on pre-specified shape protection structures (e.g., rings, linear paths, p-cycles, trees, ...). Those approaches achieve foreseeable performances and are characterized by some drawbacks related to the shapes of their protection building blocks.

In this paper, we propose new design methods to achieve different classes of quality-of-recovery independently of the shapes of the protection structures. We use shared protection schemes based on pre-configured structures (p-structures} which are either totally or partially pre-cross connected ahead of failures, or dynamically reconfigured in case of a failure to meet a specific protection need. Our classification of quality of recovery is based on the induced recovery delay in case of a link failure, and based on the required number of reconfiguration of cross connects along each backup path and at each switching node.

In order to identify the protection structures to provide the targeted quality-of-recovery classes, we use large scale optimization tools based on a decomposition method named Column Generation (CG). In our CG based optimization methods, the shape and protection capability of each protection structure is decided dynamically during the optimization process. Extensive experiments are carried out within different network scenarios and design constraints, and on different realistic network topologies in order to show the flexibility and the scalability of the proposed protection schemes and optimization methods to design them.