Multi-link failure localization via monitoring bursts

This paper introduces a novel monitoring trail (m-trail) allocation method for achieving local unambiguous failure localization under the monitoring burst (m-burst) framework, in which a single monitoring node (MN) can localize any multi-link failure with up to d links in a (d + 1)-connected network by inspecting the optical bursts traversing through it. Specifically, the proposed m-trail allocation method ensures that any healthy link is traversed by at least one uninterrupted m-trail during a multi-link failure, which can be achieved by launching no more than (d + 1)-link-disjoint m-trails per link originating from the MN. Based on such a sufficient condition, we study the performance of the m-burst framework by solving an integer linear program (ILP) and a novel heuristic, and implement the method for up to three link failures. To avoid high computation complexity in solving the ILP, a heuristic algorithm is developed for deriving (d + 1)-link-disjoint m-trails between the MN and each link of the network. Numerical results show that the proposed methods yield significantly better performance than previous methods in the reference backbone network topologies.