Spatio-temporal estimation of mobile-phone call demand in the Kumamoto earthquakes

Abstract

Due to the increasing importance of mobile communications in disaster mitigation and relief, it is very critical to be able to communicate and share critical information with others anytime and anywhere during natural disasters such as earthquakes. However, the mobile network infrastructures are usually not only physically damaged by the disasters but also congested by a tremendous traffic of mobile-phone calls. Therefore, being able to estimate the mobile-phone call demand is a key for mobile network operators to better prepare for and respond to the network congestion before and during the disasters. In this paper, we propose a data-driven approach to estimate the spatio-temporal mobile-phone call demand by leveraging the big data collected from the operational mobile network. Specifically, we develop a model that shows how the mobile-phone call demand varies with population based on the collected data from public industrial surveys and reports. We apply it to the live population distribution big data during the Kumamoto earthquakes to demonstrate the mobile-phone call demand changes spatially and temporally on the geographical map. In addition, we also demonstrate the estimation of mobile-phone call demand during the worst-case — a hypothetical earthquake occurs in the daytime. Finally, we discuss the important factors that affect of mobile-phone call demand and the trend of future mobile networks and services that have great potential to alleviate the network congestions in future disasters.