Abstract

In opportunistic networks, data is forwarded between human carried mobile devices when they are within communication range of each other (in contact). State of the art solutions to opportunistic networking problems (routing, content placement, etc.) often use the contact graph of a mobility scenario to exploit the structure of who meets whom. In this talk, we start by quantifying the structure of contact graphs with a complex network analysis. Based on insights of this analysis, we observe that current mobility models fail at correctly reproducing certain aspects of human mobility, and we present a “social overlay” model to fix this. Further, we propose a clustering-based algorithm with which nodes can infer the contact graph. Using simulation, we show that contact graph based routing protocols achieve good performance when using our algorithm, while performance degrades when constructing a contact graph naively.

This is work I have done during my PhD at ETH Zurich and presented at Infocom 2010, MobiHoc 2011 and Netscicom 2011.