Network traffic clustering and geographic visualization

Abstract

NETWORK TRAFFIC CLUSTERING AND GEOGRAPHIC VISUALIZATION by Ali Hushyar The exploration and analysis of large databases of information are an everchallenging task as digital data acquisition continues to progress. The discipline of data mining has often been employed to extract structure and patterns from the underlying dataset. In addition, new research in the field of information visualization is being applied to the same challenge. Visual models engage the invaluable pattern processing abilities of the human brain which leads to new areas of insight otherwise undetected. This research applies the benefits of both data mining and information visualization to the specific problem of traffic analysis on computer networks. This is an important issue as it relates to the ability to understand diverse behavior on the network and provide many fundamental services. For example, distinct traffic classifications and associated traffic volumes facilitate capacity-planning initiatives. Furthermore, accurate categorization of network traffic can be leveraged by quality of service offerings and, at the same time, lend itself to efficient security analysis. In this research, an example of a data processing pipeline is described that incorporates both data mining and visualization techniques to cluster network flows and project the traffic records on a geographic display.