Knowledge-based alarm surveillance for TMN

Alarm surveillance (AS) and alarm correlation (AC) are important functions in telecommunications management network (TMN) especially in heterogeneous communication networks such as public switching systems. This paper focuses on a new methodology for alarm correlation and alarm surveillance by using the deductive model-based and the inductive case-based reasoning technique. This approach is based on the object-oriented description of a network corresponding to international accepted standards and on the availability of expert knowledge about specific faults and resulting alarm situations. Long term analysis and diagnosis are supported by statistical fault hypotheses evaluation not covered explicitly by reasoning techniques. A working prototype has been successfully constructed which proved these alarm correlation principles.