Profiling Network Applications with Fuzzy C-Means Clustering and Self-Organizing Map

This paper introduces two clustering methods to be used in the analysis of network traffic. The methods are the self-organizing map (SOM) and the fuzzy c-means clustering (FCM) algorithm. First the basic theory of both methods is presented. Before the clustering process, a comprehensive data preprocessing is performed. Methods are used to produce application profiles and clusters from network traffic data. Application profiles produce significant information of the network's current state and point out similarities between different applications. This information will be later used to manage the network resources. We present a new way to compare the results obtained with SOM and FCM with enhanced nearest prototype (NP) classification algorithm. Both methods gained good results.