Two-hop privacy-preserving nearest friend searches

Nowadays, social networks are a part of everyday life. Almost everyone possessing a computing device, even a mobile one, such as a smartphone or tablet, has access to these networks. Interacting with them often requires sharing information both with the other users of the social network and with the social network itself. One of the cases that information has to be exchanged is by using services such as Facebook’s “Nearby Friends,” where a user has to share her location in order to locate her nearby friends, an action that undermines the user’s privacy. Current privacy preservation mechanisms only consider range nearest neighbor queries for nearest friend searches, limiting private friend discovery within a user’s predefined range. In this paper, we take private friend searches a step further, by presenting Two-Hop Privacy, a novel method for discovering a user’s nearest friends within arbitrary distance, not being constrained by range boundaries, in sublinear time, preserving, at the same time, the location privacy of all involved users. This is achieved by exploiting positional information of publicly available datasets of points of interest together with a randomized selection algorithm. Two-Hop Privacy is fast, requiring less than 9 ms to locate the 64 nearest neighbors between 5000 interconnected users, and capable of achieving accuracy up to 100%.