INFERRING SOCIAL NETWORKS FROM PASSIVELY COLLECTED WI-FI METADATA

Abstract

The emergence of smartphones and other highly portable Wi-Fi enabled devices offers unprecedented amounts of information leaked through Wi-Fi metadata. The constantly connected nature of Wi-Fi devices together with the intimate relationship between users and their device presents an opportunity for using a user’s device to gain information about the user themselves. Through passive data collection, without interference or the possibility of being detected, it is possible to harvest large datasets. This work looks at the possibility of inferring underlying social networks through the analysis of these metadata traces. Using spatiotemporal proximity as an indicator of friendship, findings demonstrate the ability to accurately predict underlying social structures in various simulated settings.