Using empirical mode decomposition for subscriber behaviour analysis in cellular networks in South Africa

Abstract

This paper looks at the potential benefit of using empirical mode decomposition (EMD) for the decomposition of time series data generated from a typical cellular network in South Africa. It is shown that a robust method for the extraction of features that correlate to subscriber behaviour can be conducted by decomposing time series tele-traffic data into finite set of components generated iteratively using the EMD approach. The extracted features are useful for the planning and estimation of future demand in wireless cellular networks especially in areas where subscriber socio-economic factors play a vital role in subscriber demand.