Alarm Correlation Analysis And Fault Location In Tree-like Networks With Hierarchical Structure

For the communication networks at present, the kind of tree-like hierarchical structure network is a common structure in communication networks. With the development of communication networks, the number of alarms in this kind of tree-like networks with hierarchical structure is increasing, which thus brings about the augment of the alarm database. This consequence will certainly make this issue, how to deal with the alarms and faults, a significant task to solve now. Considering the problem of low efficiency, pointlessness and the redundancy in traditional methods, we thus make a further study on this issue so as to give some solutions to handle this.Taking the disadvantages of low efficiency and pointlessness into account, we propose a kind of new method to improve these shortcomings, which is a kind of weighted Apriori algorithm based on database division. First of all, we reduce the scale of the database and make the clustering division of database based on the correlation of network nodes, so as to enhance the efficiency of data mining. Then, combined with the attributes of alarm items, the Analytic Hierarchy Process method, we further employ the weighted Apriori correlation analysis to mine frequent itemset. Finally, through our simulation, we verify that our proposed algorithm can enhance the efficiency and pertinence of the analysis of alarms and faults in tree-like hierarchical structure network.In order to further deal with the redundancy problem of fault location in tree-like hierarchical structure network, we also propose a novel analytical model for this issue. In our proposed model, we extend the conventional correlation analysis between alarms to the analysis between alarm and fault. In this paper, according to the results of the correlation analysis of alarms, we extract the useful the information and delete the redundancy, then we deduce the correlation rules between the alarm and the fault, which enhance the efficiency of fault location and accomplish the orientation of network faults. Eventually, the simulation results show that our method is easier to manage and update as well as more adaptive to the complex and instable tree-like networks with hierarchical structure.