Spatio-temporal QoS Pattern Analysis in Large Scale Internet Environment

For enhanced Quality of Service (QoS) provision of multimedia applications in Internet environment, there is a need of data mining tools supporting the automated analysis of QoS behaviour and dependencies for the purpose of modelling and forecasting, QoS planning and anomaly detection. This paper presents a data mining technology for spatio-temporal QoS pattern analysis including automated extraction and description, similarity matching and dependency maintenance of patterns in telecommunication networks. The technology is based on selection and definition of patterns from measured time series data sequences of QoS parameters using an appropriate pattern description language, pattern matching algorithms with different options for pattern similarity analysis and data mining interface supporting the network engineer in QoS pattern analysis considering temporal and spatial constraints. The building of an appropriate archive of detected similar patterns in given spatio-temporal context, i.e. pattern analysis data base, is directly concerned with for solving of specific QoS data mining task. The described technology is developed in the framework of INTERMON project for integrated QoS analysis in a large scale inter-domain environment based on interaction with monitoring, topology discovery and traffic analysis tools.