Abstract: In one aspect, a method for identifying faults in a network that has generated a plurality of fault alarms is provided. A plurality of object/alarm pairs are generated and an identification vector is assigned to each pair. The pairs are processed based on a set of rules to generate for each pair an alarm propagation vector that identifies that pair together with any other pairs from the plurality of pairs. The alarm propagation vectors are processed to determine which pairs have alarm propagation vectors that are not sub vectors of any of the alarm propagation vectors of the other pairs. Any pairs thus determined are identified as being linked to potential root causes of network problems.

Abstract: There is described a method for identifying faults in a network that has generated a plurality of fault alarms. In the method a plurality of object/alarm pairs are generated, each pair comprising a respective one of the fault alarms and a respective network object responsible for that fault alarm. A respective identification vector is assigned to each respective object/alarm pair of the plurality of object alarm pairs; and the plurality of object/alarm pairs are processed in dependence upon a set of network fault alarm propagation rules to generate for each respective object/alarm pair an alarm propagation vector that identifies that object/alarm pair together with any other object/alarm pairs from the plurality of object/alarm pairs whose alarm may potentially have arisen as a result of that object/alarm pair.