Quality of consumer experience data mining for mobile multimedia communication networks: learning from measurements campaign

This paper presents an experimental approach of the major data mining algorithms which can be applied to the quality prediction of provided services by contemporary mobile communication networks. Recent developments in performance evaluation of rollout mobile communication networks are based on drive-test measurement campaigns as well as on network monitoring and management systems. In addition, large measurement databases as well as data warehouses are created, updated and used for a drill-down statistics analysis. An empirical comparison of modern data mining methods for quality prediction and estimation of speech and video telephony services is presented. Learning prediction models are constructed from live network measurements which have been previously collected by experimental equipment during measurement campaigns. Practically and summing up, the proposed engineering approach is general and is applicable to various radio access network technologies during planning and optimisation phases.