Poster: Discovering User Relationships Through Smartphone Wi-Fi Probes

Abstract

People interact frequently with others in their daily life. Users with similar mobility patterns should have a certain degree of social relationships. Therefore, to discover user relationships, we focus on the similarity of users’ behavior patterns. Our first observation is that users with high interaction frequency are more likely to have relationships. Our second observation is that users who stay together for a long time are more likely to be related to each other, or have potential relationship. Third, we assume that users who always meet at the same place are likely to have a kind of relationships. Now, it is possible to collect data from smartphones and infer user social relationships and activities[1]. We propose a new probabilistic model for analyzing human interaction data, represented as a set of proximity links between pairs of users add with the interaction timestamp. We conduct our analysis with a slice-based approach, where all links within 10 minutes are grouped together, forming a slice of the dynamic social links graph. As shown in Figure 1, we collected raw Wi-Fi Direct proximity links over months of real life interaction to infer actual events in the life of a community. When Wi-Fi Direct devices sense the environment, they can also detect Wi-Fi access point(Red Node in Figure 1), which can be used to infer the location of users and their interactions. The time and location of the interaction are keys to deduce the interaction type.