Improving Restoration Success in Mesh Optical Networks

In shared mesh restoration, a distributed signaling protocol is used to reroute connections from failed service paths to restoration paths upon failure events. However, even when the network contains sufficient capacity, the restoration paths could be blocked for two different reasons: (1) with distributed restoration path selection schemes, multiple restoration paths may compete for the capacity of the same logical links, even when other logical links have sufficient capacity; (2) multiple restoration path set up attempts may compete for the same channels within the logical link (the "glare problem"), even when sufficient capacity is available within the logical link. This paper proposes a hybrid distributed/centralized restoration mechanism for restoration path selection and a channel selection scheme that eliminates almost all glares. As shown from simulation in a typical intercity backbone network, our proposed hybrid mechanism improves the first restoration attempt success ratio by 40% compared to a pure distributed restoration approach. In addition, the proposed restoration path selection algorithm saves up to 50% of restoration capacity compared with the disjoint shortest path algorithm and 15% compared with a previously published greedy algorithm.