Distributed WiFi detection and integration in dense urban mobile Peer-to-Peer networks

Running Peer-to-Peer applications??such as multimedia streaming or file sharing??on mobile devices significantly increases the congestion in 3G access networks. Offloading traffic from 3G to WiFi domains is promising in such scenarios, since communication is possible without generating any load in the WiFi??s uplink or in the Internet, given that peers are located in the same WiFi domain. However, in today??s urban areas devices are commonly in range of dozens of infrastructure-based WiFi domains, a fact that calls for an efficient rendezvous mechanism. In this article, we propose a rendezvous mechanism that efficiently enables physically close mobile devices running an arbitrary P2P application to peer with each other in a common WiFi domain. The mechanism builds upon tree-based collection, aggregation, and distribution of WiFi information. Using a stochastic model, we estimate the overhead of the mechanism based on WiFi density statistics from real world urban areas. We further show how to reduce this overhead on the expense of a reduced rendezvous success probability by applying Bloom Filters. Simulations of a tree-based Peer-to-Peer media streaming application demonstrate that the mechanism can in fact support effective offloading of P2P traffic to WiFi domains.