MobiCom Poster Abstract: Super-diffusive Behavior of Mobile Nodes from GPS Traces
Sung‐Won Kim, Chul‐Ho Lee, Do Young, Eun · 2007
Mobility is the most important factor in mobile ad-hoc networks (MANETs) and delay-tolerant networks (DTNs). Mobility models that fail to capture real movement pattern of mobile nodes will result in misleading guidelines on the design of new protocols and their performance evaluations and thus prevent us from making judicious decisions. Two different approaches – synthetic modeling and tracebased modeling – have been put forth for sound mobility models. Synthetic mobility models [7] have been developed mainly for the purpose of simplicity and the ease of analysis, but subsequently been criticized for their unrealistic behaviors. On the other hand, the trace-based approach suffers from lack of the amount of available traces on a fine time/space scale, and are applicable only for the particular setting under consideration and highly sensitive on the choice of the reconstruction algorithm. In this work, we take a different approach from the above two. We first investigate numerous GPS-based mobility traces for various mobile nodes. In view of Data MULEs or Message Ferrying based approaches, we consider different kinds of human movement patterns such as walking, running, inline skating and bicycling as well as zebra and bus movement. We then find out that all those traces show super diffusive movement pattern, which is characterized by ‘faster-than-linear’ growth curve of the mean square displacement (MSD), i.e., E{‖Zt − Z0‖} ∼ t with γ > 1, where Zt ∈ R is the position of the mobile node at time t. Based on the observed super-diffusive behavior, we propose to use a set of Levy walk model as simple, easy-togenerate, yet realistic mobility models. The Levy walk models allow us to directly control the super-diffusive behavior of generated trajectories and can be easily incorporated into performance evaluation of protocols of MANETs and DTNs.