Fuzzy temporal reasoning model for event correlation in network management

Abstract

Fault management is used to detect, isolate, and repair problems in communication networks. Alarms are considered an external manifestation of faults occurring inside the managed network. These faults may affect the network's hardware and software components. Alarm correlation identifies the relationships between reported alarms to produce a new, smaller list of alarms. This allows operators to quickly and more accurately to identify the root cause of problems and resolve them faster. Temporal reasoning, reasoning about time, plays a critical role in monitoring network alarms. In this paper, a fuzzy-logic model is proposed for finding the temporal relation between events for correlation purposes. A fuzzy-inference rule base is used to integrate fuzzy membership functions to determine the "best" temporal relation between events. The proposed model is compared to the traditional crisp temporal reasoning.