The study of alarm association rules mining in telecommunication networks

Association rules mining plays an important part in the alarm correlation analysis in the telecommunication networks. A novel algorithm based on time window pre-processing and the weighted frequent pattern tree method was proposed in this paper. It is an efficient algorithm which can avoid scanning the database many times and producing a large number of conditional pattern trees. Experiments on a large alarm data set show that the approach is practical for finding frequent patterns in the alarm correlation analysis, and the performance of WFP method is better than the classical FP-growth algorithm.