A Distributed Network Alarm Correlation Analysis Mechanism for Heterogeneous Networks

Abstract

The current China State Grid network consists of a set of heterogeneous and distributed networks. In such a heterogeneous environment, it is a challenging task to provide efficient network alarm management, and further to analyze the correlation of alarms. Thus, an efficient and distributed network alarm analysis scheme is becoming necessary and indispensable. To this end, we have attached our emphasis on two aspects of the aforementioned problem and proposed the corresponding algorithms, respectively. Firstly, we introduced an intra-network alarm treatment process to manage the heterogeneous network alarms. Especially, we leverage on the fuzzy clustering method and fuse alarms within the same cluster into comprehensive alarms by utilizing the Dempster–Shafer theory. Secondly, we proposed an inter-network alarm analysis process to mine the correlation rules of alarms through a distributed scheme based on the Frequent Pattern-Growth (FP-Growth) association rule mining algorithm. Compared with the traditional centralized scheme and another Apriori-based distributed algorithm, our proposed scheme has a higher time efficiency for the effective management of network alarms. With the aid of such a two-step network alarm management scheme, it is easy for the network management system to make a better management of alarms in heterogeneous and distributed networks.