A survey of network faults classification using clustering techniques

With the rapid development of networking applications of different kinds, the network administrators need highly effective tools to classify faults that occur in the networks. The real time classification of such faults can help manage networks efficiently. In our previous work we studied clustering algorithms to classify network faults and provide an evaluation based on the speed, accuracy, efficiency and cost. Out of them Fuzzy Cluster-Means is considered more suitable for network faults classification. In this paper we implemented FCM in MATLAB to classify network faults. We built a prototype to demonstrate the efficiency of the algorithm. The empirical results are compared with other algorithms such as Neural Networks. Results reveal that FCM outperforms other algorithms. Its visualization of classified faults can help network personnel to take well informed decisions pertaining to security.