Multi-layer sociality in opportunistic networks: An extensive analysis of online and offline node social behaviors

The large diffusion of mobile devices able to derive human sociality from wireless encounters and the growing use of online social networks is increasingly driving the research community on opportunistic networks towards social-based techniques and hence, towards the analysis of user social behavior. Within these challenged networks where node connectivity is highly intermittent and contact opportunities are exploited to communicate without network infrastructure, node mobility is basically driven by human sociality. As such, understanding the social behavior of nodes within these networks is of paramount importance, especially for finding suitable relays in message forwarding. In this paper, we focus on the analysis of a collection of multi-layer social networks derived from six different datasets containing mobility data and Facebook friendships of nodes moving in opportunistic network environments. Analyzing egocentric and sociocentric node behaviors on the opportunistic social network detected through wireless encounters and on the corresponding Facebook social network, we show that online and offline degree centralities are significantly correlated on most datasets. On the contrary, betweenness, closeness and eigenvector centralities show medium-low correlation values. Considering that opportunistic networks are highly dynamic and in the bootstrapping phase of the network having a clear social behavior of nodes in terms of offline centrality is one of the main issues, these results show that in some cases, online centrality which is easier to compute can predict offline node centrality. Moreover, we show that in most datasets, most of the strong ties correspond to Facebook friendships.