Fine-grained cloaking of sensitive positions in location sharing applications

Geo-social networking applications magnify the concern for location privacy because the users' positions are disclosed to several parties and not all of these parties are trusted. A challenging privacy requirement concerns the protection of sensitive positions against the untrusted members of a geo-social network. Consider the case of a user who does not want to reveal her friends the sensitive place in which she is located, e.g., hospital. Typically the user discloses a cloaked region in place of the exact position. However a malicious observer aware of the geography of the territory as well of the distribution of positions can easily defeat the protection mechanism and detect the place in which the user is located. In this paper we elaborate on this inference attack, named semantic location identi cation, and present a comprehensive framework supporting semantic location cloaking.