Intelligent search for correlated alarm events in databases

The main topic of this paper is fault management, especially the search for correlated alarms in large alarm records stored in databases. We have chosen a GSM/DCS mobile telephone network as a basis for the investigations of our data mining algorithms which are used to detect correlated alarm patterns. Alarm correlation tools need information for the correlation of alarms which today are only known to system experts. The fallibility of human operators was the motivation to develop a data mining tool that is able to find correlated alarms. The parts introduced in this paper describe the main ideas of the algorithms and explains where the detection process is optimised. The algorithms implemented in our data mining tool ISIS (intelligent search of interesting patterns in sequences) are introduced and the functionality is explained