Mining Time Series Data in Mobile Cellular Networks
Anish Matthew Kurien, Barend Jacobus van Wyk, Yskander Hamam · 2008
According to the ITU [2], the world population in 2006 amounted to 6.6 billion inhabitants, out of which, 923 million lived in Africa. The number of mobile subscribers sat at 198 million which amounted to nearly 7.2% of the worldwide mobile subscribers. Mobile telephony has been viewed as a critical enabling technology that is capable of boosting economies across Africa. However, Africa accounted for only 14% of the worldwide Gross Domestic Product (GDP) in 2006 [2]. With the varying socio-economic distributions that is prevalent in most African countries, an accurate mechanism that is able to determine traffic trends in a mobile cellular network based on subscriber behaviour would be beneficial to an operator for planning of network demand. With the availability of large amounts of data from existing networks, data mining techniques that are able to retrieve meaningful information that is beneficial to a network planner would be useful. This paper looks at the benefits of using data mining in time-series databases for the determination of traffic trends in mobile cellular networks.