Self-Organizing Networks (SON) aim to automate network Operation & Maintenance tasks. SONs comprise self-configuration, self-optimization and self-healing. Within self-healing, root cause analysis, i.e. diagnosis of the cause of the network problems, is one of the most difficult tasks. To automate diagnosis, Data Mining (DM) algorithms over sets of solved troubleshooting cases can be applied in order to use Knowledge Based Systems. Data reduction is part of the DM process, where large time-dependent matrices of Performance Indicators (PIs) are transformed into time-independent vectors of values. In this work, an algorithm for data reduction is proposed, which is based on detecting the time intervals when the service of an LTE eNodeB is degraded and aggregating the values of the time dependent PIs for those intervals. The results show that the detecting capability of the algorithm is higher than other proposed solutions, and that a high volume reduction factor can be achieved.
