A New Efficient Clustering Algorithm for Network Alarm Analysis

In this paper, we introduce an efficient approach for modelling the problem of faults’ identification and alarms correlation in large telecommunication networks. These alarms are usually useful for identifying faults in such systems, and therefore solving them. However, for large systems, the number of alarms produced is so large that the current management systems cannot handle it. For instance, a single fault may result in a large number of alarms, and it is often very difficult to isolate the true cause of the fault. One way of overcoming this problem is to analyse and interpret these alarms before faults can be located. Alarms are triggered with a particular period while the fault still remains, so we need an efficient algorithm to identify sets of alarms related to a particular fault. In this paper we present a general algorithm for identifying faults by clustering alarms according to their identification and periodicity. An evaluation on real data from a commercial network is presented and discussed.