You Can Act Locally With Efficiency: Influential User Identification in Mobile Social Networks

As mobile social networks grow rapidly, influential user identification has attracted much more attention. Previous studies either need large message overhead to achieve global maxima in influence computation or focus on relatively stable network topology. To tackle the dynamic topology, we present an influential user identification scheme that fully exploits the active mobile users, in which the stable-state property could be leveraged under information potential construction scheme. We also propose an efficient routing algorithm for reaching the global maxima without depending on specific routing protocols. The proposed scheme is validated with extensive simulations using both synthetic random-walk and real-world mobility traces. The results demonstrate that it achieves considerable performance on influential user identification and route construction with little overhead. Furthermore, we present a case of mobile data offloading, and the results show that our scheme could reduce the efficient data traffic by up to 79.2%, compared with a baseline without data offloading.