Epidemic forwarding in mobile social networks

Recent years have witnessed the prosperity of mobile social networks, where various information is shared among mobile users through their opportunistic contacts. To investigate efficiency of information dissemination in wireless networks, epidemic models have been employed to study message forwarding delays, presuming message delivery whenever an opportunistic contact occurs. A practical concern is typically neglected, that one mobile user may only be willing to pass information onto others with social ties, rather than anyone upon contact. Under such a constraint, information dissemination may behave differently, according to the pattern of social ties that exist in the network. In this paper, we model social-aware epidemic forwarding in mobile social networks using mean-field equations, and carefully study the end-to-end unicast message propagation delays under different levels of social ties among users. Both cases of limited and unlimited message validity are considered in our models, i.e., whether relay nodes may delete a message after carrying it for some finite time T or never. Through careful theoretical analysis and empirical studies, we made a number of intriguing observations: First, the topology of social relation graphs significantly influences message forwarding delays, i.e., the more skewed the social relationship distribution is, the larger delay it results in. Second, the average delivery delay remains fairly stable with the growth of system scale, presenting a sharp contrast with the case without social awareness. Third, we observe that with a moderate choice of T, message delivery can achieve a successful ratio of almost 100% with an expected delay very close to the case of unlimited validity, signifying that a good tradeoff can be achieved between end-to-end message delivery efficiency and energy/storage overhead at the relay nodes in a network. All these provide useful guidance for efficient information dissemination protocol design in practic- l mobile social networks.