Research On Faulty Detection And Localization In All-optical Networks

Fault management is a crucial functional block in all-optical networks,all services on the link will be disrupted once the fiber is destoried.Therefore it is necessary to localize the exact location for failure diagnose when the failure occurs to restore the disrupted services for further failure isolation and repairment.Due to the increasing applications and bigger traffic load in the network,minimum probing to diagnose network failures is of great concern where excessive resource consumption,a key attention in non-adaptive scheme,need to be solved.Only if fiber link failure can be quickly monitored and localized,each layer can accordingly adjust routing protocol so as to timely recover data transmission by transferring data on the failed link to other links.So we are endeavor to solve the problem of the long diagnose time in adaptive fault localization.In order to solve these problems,the works of this paper are shown as follows:1.Againist adaptive multi-link failure localization problem,this paper studies the probe selection problem in all-optical networks for achieving unambiguous faulty links localization.Existing random walk algorithm just find feasible solutions to localize the faulty link unambiguously,but it consumes large number of probes and wavelengths in large-size networks.Firstly,monitoring paths are set up for failure detection followed by sending probing signals on all monitoring paths simultaneously,and then failure localization is performed on the failed monitoring paths.The proposed scheme proved the problem of minimal monitoring path set is complete NP problem and found it through heuristic monitoring path selection algorithm.Meanwhile,the new scheme also proved that the sufficient and necessary condition of localizing k faulty links with one monitoring node is the network must be k(10)1 edge connectivity.Compared with random walk algorithm,probe selection algorithm greatly shortens the number of probe and consumed wavelength per link.2.The non-adaptive approach for quickly localizing faulty links is proposed in this paper owing to long diagnose time in adaptive fault localization.First of all,we develop a binary observation model in which end-to-end observations are presented with individual link status.Secondly,a faulty links localization algorithm originally developed in group testing is adopted to identify the locations of the faulty links.Simulations indicate that a non-adaptive faulty links localization algorithm can localize each faulty link in a short time as well as lead to a significantly lowered computational complexity.