On Integrating Failure Localization with Survivable Design

Abstract

In this thesis, I proposed a novel framework of all-optical failure restoration which jointly determines network monitoring plane and spare capacity allocation in the presence of either static or dynamic traffic. The proposed framework aims to enable a general shared protection scheme to achieve near optimal capacity efficiency as in Failure Dependent Protection(FDP) while subject to an ultra-fast, all-optical, and deterministic failure restoration process. Simply put, Local Unambiguous Failure Localization(L-UFL) and FDP are the two building blocks for the proposed restoration framework.

Under L-UFL, by properly allocating a set of Monitoring Trails (m-trails), a set of nodes can unambiguously identify every possible Shared Risk Link Group (SRLG) failure merely based on its locally collected Loss of Light(LOL) signals. Two heuristics are proposed to solve L-UFL, one of which exclusively deploys Supervisory Lightpaths (S-LPs) while the other jointly considers S-LPs and Working Lightpaths (W-LPs) for suppressing monitoring resource consumption. Thanks to the ``Enhanced Min Wavelength Max Information principle'', an entropy based utility function, m-trail global-sharing and other techniques, the proposed heuristics exhibit satisfactory performance in minimizing the number of m-trails, Wavelength Channel(WL) consumption and the running time of the algorithm.

Based on the heuristics for L-UFL, two algorithms, namely MPJD and DJH, are proposed for the novel signaling-free restoration framework to deal with static and dynamic traffic respectively. MPJD is developed to determine the Protection Lightpaths (P-LPs) and m-trails given the pre-computed W-LPs while DJH jointly implements a generic dynamic survivable routing scheme based on FDP with an m-trail deployment scheme. For both algorithms, m-trail deployment is guided by the Necessary Monitoring Requirement (NMR) defined at each node for achieving signaling-free restoration. Extensive simulation is conducted to verify the performance of the proposed heuristics in terms of WL consumption, number of m-trails, monitoring requirement, blocking probability and running time.

In conclusion, the proposed restoration framework can achieve all-optical and signaling-free restoration with the help of L-UFL, while maintaining high capacity efficiency as in FDP based survivable routing. The proposed heuristics achieve satisfactory performance as verified by the simulation results.