Understanding human mobility and its impact on wireless mobile networks

Abstract

In this thesis, we study the human mobility to build up essentially realistic mobility models. Through our own large scale GPS experiments carried on 5 sites in the world by over 100 volunteers, in the first part, we identify the statistical properties of human walks and verify that the mobility of humans closely resembles Levy-walk known as an animal foraging pattern. Based on the observation, we present a truncated Levy-walk mobility model and evaluate the performance of wireless mobile networks including MANETs and DTNs on our mobility model. In the second part, we further study the Levy-walk characteristics of humans to demystify the underlying mech anisms and embed the correlation among humans to our mobility model. We first identify that the 2 dimensional distribution of contexts has self-similarity resulting fractal property. Over the given contexts, by proposing a least action trip planning algorithm (LATP) mimicking a human decision mechanism on movement, we present a new mobility model SLAW (self-similar least action walk) to reproduce the Levy-walk of humans sharing common contexts. In third part, we identify the network scaling performance over the mobility models that we suggested. We focus on the delay and capacity trade-off of wireless mobile networks with Levy-walk mobility and clarify the trade-off regions for the mobility. Combining with the previous work [1], we complete the global perspective of the delay and capacity trade-off for remarkable mobility models covering i.i.d. mobility, random direction mobility, Brownian mobility and Levy-walk mobility in wireless mobile networks.