From Human Routine to More Efficient Mobile Networks

Abstract : The proliferation of pervasive communication caused a recent boost up on the mobile data usage, which network operators are not always prepared for. The main origin of the mobile network demands are smartphone devices. From the network side those devices may be seen as villains for imposing an enormous traffic, but from the analytical point of view they provide today the best means of gathering users information about content consumption and mobility behavior on a large scale. Understanding users' mobility and network behavior is essential in the design of efficient communication systems. We are routinary beings. The routine cycles on our daily lives are an essential part of our interface with the world. Our habits define, for instance, where we are going Saturday night, or what is the typical website for the mornings of Monday. The repetitive behavior reflects on our mobility patterns and network activities. In this thesis we focus on metropolitan users generating traffic demands during their normal daily lives. We present a detailed study on both users' routinary mobility and routinary network behavior. As a study of case where such investigation can be useful, we propose a hotspot deployment strategy that takes into account the routine aspects of people's mobility. We first investigate urban mobility patterns. We analyze large-scale datasets of mobility in different cities of the world, namely Beijing, Tokyo, New York, Paris, San Francisco, London, Moscow and Mexico City. Our contribution is this area is two-fold. First, we show that there is a similarity on people's mobility behavior regardless the city. Second, we unveil three characteristics present on the mobility of typical urban population: repetitiveness, usage of shortest-paths, and confinement. Those characteristics undercover people's tendency to revisit a small portion of favorite venues using trajectories that are close to the shortest-path. Furthermore, people generally have their mobility restrict to a dozen of kilometers per day. We then investigate the users' traffic demands patterns. We analyze a large data set with 6.8 million subscribers. We have mainly two contributions in this aspect. First, a precise characterization of individual subscribers' traffic behavior clustered by their usage patterns. We see how the daily routine impacts on the network demands and the strong similarity between traffic on different days. Second, we provide a way for synthetically, still consistently, reproducing usage patterns of mobile subscribers. Synthetic traces offer positive implications for network planning and carry no privacy issues to subscribers as the original datasets. To assess the effectiveness of these findings on real-life scenario, we propose a hotspot deployment strategy that considers routine characteristics of mobility and traffic in order to improve mobile data offloading. Carefully deploying Wi-Fi hotspots can both be cheaper than upgrade the current cellular network structure and can concede significant improvement in the network capacity. Our approach increases the amount of offload when compared to other solution from the literature.
Complete list of metadatas
Contributor : Eduardo Mucelli Rezende Oliveira <>
Submitted on : Thursday, June 4, 2015 - 8:32:30 PM
Last modification on : Tuesday, February 5, 2019 - 2:38:01 PM
Long-term archiving on : Tuesday, April 25, 2017 - 3:17:27 AM




  • HAL Id : tel-01160280, version 1



Eduardo Mucelli Rezende Oliveira. From Human Routine to More Efficient Mobile Networks. Networking and Internet Architecture [cs.NI]. École Polytechnique, 2015. English. ⟨tel-01160280⟩



Record views


Files downloads