Towards connecting people, locations and real-world events in a cellular network

The success of personal mobile communication technologies has led an emerging expansion of the telecommunication infrastructure but also to an explosion to mobile broadband data traffic as more and more people completely rely on their mobile devices, either for work or entertainment. The continuously interaction of their mobile devices with the mobile network infrastructure creates digital traces that can be easily logged by the network operators. These digital traces can be further used, apart from billing and resource management, for large-scale population monitoring using mobile traffic analysis. They could be integrated into intelligent systems that could help at detecting exceptional events such as riots, protests or even at disaster preventions with minimal costs and improve people safety and security, or even save lives. In this paper we study the use of fully anonymized and highly aggregate cellular network data, like Call Detail Records (CDRs) to analyze the telecommunication traffic and connect people, locations and events. The results show that by analyzing the CDR data exceptional spatio-temporal patterns of mobile data can be correlated to real-world events. For example, high user network activity was mapped to religious festivals, such as Ramadan, Le Grand Magal de Touba and the Tivaouane Maouloud festival. During the Ramadan period it was noticed that the communication pattern doubled during the night with a slow start during the morning and along the day. Furthermore, a peak increase in the number of voice calls and voice calls duration in the area of Kafoutine was mapped to the Casamance Conflict in the area which resulted in four deaths. Thus, these observations could be further used to develop an intelligent system that detects exceptional events in real-time from CDRs data monitoring. Such system could be used in intelligent transportation management, urban planning, emergency situations, network resource allocation and performance optimization, etc.

© 2016. This author's accepted manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/