Using Monitoring Cycles for Efficient Multi-Link Failure Diagnosis on All-Optical Mesh Networks

Abstract

Recently, network failure diagnosis usually adopted simple topologies as their experiments. Yet, the limitation of one -way transmission on the optical fibers is often overlooked. For this reason, our research assumes two directed optical fibers, which are in opposite directions to each other, are placed to connect any two neighboring nodes in an all-optical mesh network. Thus, the network architecture formed in this manner can be viewed as a bidirectional topology which better reflects the configuration facts and the implementation requirements. According to the above assumption, a monitor selection method considers the condition of bi-directional topology is proposed in our work first. In this method, fewer monitor activation with the same fault diagnostic capability becomes an important objective so that the concept of health information is introduced to provide additional networks status for the aids of failure diagnosis. Moreover, to increase the practicability of our mechanism, our work further adapt the mechanism to more complex networking environments and situations, e.g., occurrence of multiple simultaneous faults on a metro-core mesh networks. Finally, the experimental results are compared with various methods, like as m-cycle, m^2-cycle, redundant monitor deactivation problem (RDMP), and parallel limited perimeter vector matching (P-LVM), and the results reveal that our mechanism can function equally well even with large and complex topologies.