Group shared protection for spare capacity reconfiguration in optical networks

摘要

This paper introduces GSP, a group shared protection scheme, for wavelength division multiplexing (WDM) mesh networks with dynamically arrived connection requests. Based on the (M:N)~n control architecture, GSP has n mutually independent protection groups (PGs), each of which containing N shared risk link groups (SRLG) disjoint working paths protected by M protection paths. Due to the SRLG-disjointedness of the working paths in each PG, GSP not only allows the spare capacity to be totally sharable among the corresponding working paths, but also reduces the number of working paths affected due to a single link failure. Based on the framework, an integer linear program (ILP) formulation that can optimally reconfigure the spare capacity for a specific PG whenever a working-protection path-pair joins is proposed. This approach is appropriate for a dynamic traffic scenario where inter-arrival time is large and where arriving request can tolerate some delay, but may not be suitable where traffic arrival rate is high and incoming requests need to be served within a few seconds. To trade the performance (i.e., capacity efficiency) with the computation complexity, two heuristics, namely ring-shared protection (RSP) and link-shared protection (LSP) are proposed. The proposed schemes are compared with an exiting one, namely the successive survivable routing (SSR). The simulation results show that LSP, RSP and SSR yield similar performance in terms of resource sharing, whereas ILP outperforms all of them by 6-16%. Due to the limited number of working paths in each PG, ILP can handle dynamically arrived connection requests in a reasonable amount of time. Also, we find that the number of affected working paths in GSP is about half of that in SSR. We conclude that GSP provides a scalable and efficient solution for dynamic spare capacity reconfiguration following the (M:N)~n control architecture.