SocialSwarm: Exploiting distance in social networks for collaborative flash file distribution

Social networks can serve as an effective mechanism for distribution of vulnerability patches and other malware immunization code. We propose a novel approach - SocialSwarm - by which peers exploit distances to their social peers to approximate levels of altruism and to collaborate on flash distribution of large files. SocialSwarm supports heterogeneous BitTorrent swarms of mixed social and non-social peers. We implement SocialSwarm as an extension to the Rasterbar libtorrent library - widely used by BitTorrent clients - and evaluate it on a testbed of 500 independent clients with social distances extracted from Facebook. We show that SocialSwarm can significantly reduce the average file distribution time, not only among socially connected peers, but also among other swarm participants.