| Description |
We develop a novel framework for friend-to-friend (f2f) distributed services (F3DS) by which applications can easily offer peer-to-peer (p2p) services among social peers with resource sharing governed by approximated levels of social altruism. Our frame- work differs significantly from typical p2p collaboration in that it provides a founda- tion for distributed applications to cooperate based on pre-existing trust and altruism among social peers. With the goal of facilitating the approximation of relative levels of altruism among social peers within F3DS, we introduce a new metric: SocialDistance. SocialDistance is a synthetic metric that combines direct levels of altruism between peers with an altruism decay for each hop to approximate indirect levels of altruism. The resulting multihop altruism levels are used by F3DS applications to proportion and prioritize the sharing of resources with other social peers. We use SocialDistance to implement a novel flash file/patch distribution method, SocialSwarm. SocialSwarm uses the SocialDistance metric as part of its resource allocation to overcome the neces- sity of (and inefficiency created by) resource bartering among friends participating in a BitTorrent swarm. We find that SocialSwarm achieves an average file download time reduction of 25% to 35% in comparison with standard BitTorrent under a variety of configurations and conditions, including file sizes, maximum SocialDistance, as well as leech and seed counts. The most socially connected peers yield up to a 47% decrease in download completion time in comparison with average nonsocial BitTorrent swarms. We also use the F3DS framework to implement novel malware detection application- F3DS Antivirus (F3AV)-and evaluate it on the Amazon cloud. We show that with f2f sharing of resources, F3AV achieves a 65% increase in the detection rate of 0- to 1-day-old malware among social peers as compared to the average of individual scanners. Furthermore, we show that F3AV provides the greatest diversity of mal- ware scanners (and thus malware protection) to social hubs-those nodes that are positioned to provide strategic defense against socially aware malware. |
