Detecting dynamic communities in opportunistic networks

In opportunistic networks, communities of mobile entities may be utilized to improve the efficiency of message forwarding. However, identifying communities that are dynamically changing in mobile environment is non-trivial. Based on random walk on graphs, in this paper we present a community detection algorithm that takes into account the aging and weight of contacts between mobile entities. Our idea originates from message-forwarding operations in opportunistic networks. We evaluate the algorithm on both computer-generated networks and real-world human mobility traces. The result shows that our proposed algorithm can find the communities and detect the changes in their structures over time.