A Study Of The Fault Localization Algorithm Of The Next Generation Smart Optical Network

With the increase of Internet service and development of wavelength Division Multiplexing(WDM) technique, the amount of information transferred in Smart Optical Network becomes much larger. A Single fault may trigger many alarms and cause large lost of the information in network. The quick and exact localization of the fault is very important to the whole optical network.When we are studying fault management, it is important to know the faults that may occur and to localize the faults exactly. Only in this way we can take more effective measures according to different faults so as to improve the validity and security of the whole network.As to the next generation Smart Optical Network, any fault will cause the drop of the QoS. For example, it will cause alarm of hard fault because of events that suddenly interrupt established channels or alarm of soft fault because of events that progressively degrade the quality of transmission, or both. And then these alarms will be sent to the network management system to localize the faults. But network management system will not receive the alarms quickly and exactly for the communication failure, which cause false and missing alarm, then cause the drop of network localization precision.In allusion to the above problems, the content of this thesis is as follows:Firstly, the common alarm correlation technique is analyzed and compared, then bring forward a method based on channel coding to analyze the correlation between the network device. Secondly, analyze the alarm capability of the device in network, then use the method based on binary tree to localize multiple fault. This method solve the problem of multiple hard/soft fault localization. It can take missing alarm and false alarm into account at the same time, and then improve the precision of fault localization. Thirdly, propose the localization method based on alarm correlation coding. This method uses less memory space, but can localize multiple faults more quickly. At last, the algorithm is simulated and the performance is analyzed.