Optimizing probe selection algorithms for fault localization

Due to the increasing development of communication networks, higher network availability and reliability is required. Hence, fault management is becoming an important and difficult issue in dynamic network environment. In this paper, we mainly use active probing to perform fault detection and diagnosis tasks. Past research on probing-based monitoring technique provides a preplanned probing technique which leads to extremely high computational complexity and involves potentially huge volumes of traffic. A novel method called online probing overcomes the drawbacks mentioned above. However, its time complexity is intolerable. In this paper, we propose an effective tool combining the advantages of online and preplanned probing. By analysis the features of suspected faults and narrow the scope of the suspected fault set, we can locate the root of faults in the network efficiently. Experiments show that our new approach can perform well in probe selection and has advantage of real-time and economic.