Monitoring trail (m-trail) has been proposed as an effective approach for link failure localization in all-optical wavelength division multiplexing (WDM) mesh networks. Previous studies in failure localization rely on alarm dissemination via control plane signaling such that the network controller can collect the flooded alarms to form an alarm code for failure identification. Such cross-layer signaling effort obviously leads to additional control complexity. This paper investigates a novel m-trail failure localization scenario, called network-wide local unambiguous failure localization (NWL-UFL), where each node can perform UFL based on locally available on-off state of traversing m-trails, such that alarm dissemination in the control plane can be completely avoided. The paper first defines and formulates the m-trail allocation problem under NWL-UFL and conducts a series of bound analysis on the cover length required for localizing any single-link failure. This is the first study on monitoring trail allocation problem that aims to gain understanding on the consumed cover length via analytical approaches due to the special feature of the NWL-UFL scenario. A novel heuristic algorithm based on random spanning tree assignment (RSTA) and greedy link swapping (GLS) is developed for solving the formulated problem. Extensive simulation on thousands of randomly generated network topologies is conducted to verify the proposed scheme by comparing it to a naive counterpart and with the derived lower bounds. We also demonstrate the impact of topology diversity on the performance of the proposed scheme as well as its scalability regarding network sizes.
ASJC Scopus subject areas
- Electrical and Electronic Engineering
- Computer Science Applications
- Computer Networks and Communications