Multi-fractal analysis of IP-network traffic based on a hierarchical clustering approach

This paper describes an analysis of IP-network traffic in terms of the time variations in multi-fractal scaling properties. To obtain a comprehensive view in assessing IP-network traffic conditions, we used a hierarchical clustering scheme, which provides a way to classify high-dimensional data into a tree-like structure. Based on sequential measurements of IP-network traffic at two locations, we checked time variations in multi-fractal-related properties of measured data sets. In performing the hierarchical clustering-based analysis, we used four parameters: the highest value and the range of generalized fractal dimensions, the network throughput, and the standard deviation of average throughput for each measured data set. The results confirmed that the traffic data could be classified in accordance with the network traffic properties, demonstrating that the combined depiction of the multi-fractal-related properties and other factors can give us an effective assessment of network conditions at different times.