The Research Of Fault Diagnosis Techniques In Information Networks

Fault diagnosis is a vital task for network management and it is a process of detecting a fault and localizing the root cause of the fault from the observed fault indication. Trough the summarize of fault diagnosis techniques in recent 10 years.We find that although there are many fault diagnosis techniques including methods based on detection, graph theory,and clustering,some issues remain unresolved. Most of the fault diagnosis algorithm diagnosed the network connectivity,while few algorithm diagnosed poor link quality or frequent packet loss caused by link error,congestion and other causes.And for the detection of disconnected nodes of Ad Hoc networks,most of the method are based on sending packets,there is no full use of the network of geographic information and the accuracy is not high. In this paper, these two aspects are explored and improved in this paper, including:(1)Algorithm of dynamic detecting of disconnected node in AdHoc Network is improved.This paper studied two situation, including only detecting the state of one node and detecting the of multi-nodes. Detecting whether nodes int the network is lost or not according to the node's location and communication range and reducing the complexity of the algorithm through the sort of nodes according to their location. Finally we implement this algorithm based on Opnet.According to the experimental results, the proposed algorithm is more accurate than other two algorithm.(2)The diagnosis of the link of poor link quality is studied.In this paper, delay is used to determine the link quality and probes is used to get the link delay.The standard of delay is given to determine when a is probe failed. In this paper, the adaptive probe method is used to diagnose faulty link.First,selecting some probe to detect whether a network failure occurs, then starting the fault location phase, further selecting probes to localize the failed links. Finally, the algorithm is implemented based on Opnet, the results show that this method can localize the the link of poor link quality with less probes.