Shared-Protection Survivable Multipath Scheme in Flexible-Grid Optical Networks Against Multiple Failures

Compared with fixed-grid optical network, flexible-grid optical network (FON) enabling technologies could provide higher spectrum efficiency and thus more capacity. With more data capacity, physical network failure (i.e., link or node) will lead to more communication disruptions, especially for multiple failures. In previous research, the protection strategies against multiple failures is mainly single-path provisioning (SPP). They require multiple diverse routes and allocate protection resource separately, which are resource consuming. With the advantages of partial protection and lower resources consumption, the multipath provisioning (MPP) scheme has been considered in traditional wavelength-division multiplexing network, where traffic is split into multiple lower-rate streams and each stream is separately routed. In MPP, the number of required paths and the amount of resource allocated for each path are the two important parameters to be determined and will have great impact on the spectrum efficiency. In light of the above discussion, a spectrum-efficient shared protection MPP (S-MPP) scheme against multiple failures in FONs is proposed. It is formulated as an Integer Linear Programming (ILP) model and a heuristic algorithm is also developed. Also, the determination of the number of paths and the resource requirement of each path in FON is analyzed in this paper. Furthermore, with MPP, the shared protection in FON is different from that with SPP, because the paths are equal in MPP (no primary path or backup path) and sharing partial path is available. Therefore, we discuss the partial sharing scheme and analyze the amount of shared spectrum resource on each path. Finally, the simulation results show that the proposed S-MPP scheme in FON achieves a significant gain on spectrum efficiency with enough protection against multiple failures. Moreover, the performance of the proposed heuristic algorithm is very close to the optimal solution achieved by solving ILP.