Mobility trace breeding

Abstract

Collecting real contact traces in disruption-tolerant networks is a complex procedure. From the experiment setup to the data cleaning, a lot of efforts is required. We propose a breeding system to derive possible contact traces from a single real experiment. We check the consistency of our system using synthetic mobility traces and show that bred traces do follow characteristics of the original trace. We apply then the system to a real-world contact dataset. Our results indicate that even when using incomplete or imperfect real contact traces, our system produces accurate traces.