Nowadays telecommunication systems are growing more and more complex, generating huge amount of alarms that cannot be effectively managed by human operators. The problematic is to detect in real-time significant combinations of alarms that describe an issue. In this article, we present a powerful heuristic algorithm that constructs alarm patterns dependency graphs. More precisely, it is able to highlight patterns extracted from an alarm flow learning process with a small footprint on network management system performance. This algorithm is first relevant to real-time issues detection by effectively delivering their concise alarm patterns. And secondly it allows the proactive analysis of network health by retrieving the general trends of a network. We challenge our algorithm to an optical network alarms data set of an existing operator. We find immediately similar results to the experts analysis performed for this operator by Alcatel-Lucent Customer Services.
