Server-Aided Privacy-Preserving Proximity Testing

Licentiate thesis, 2022

Proximity testing is at the core of many Location-Based online Services (LBS) which we use in our daily lives to order taxis, find places of interest nearby, connect with people. Currently, most such services expect a user to submit his location to them and trust the LBS not to abuse this information, and use it only to provide the service. Existing cases of such information being misused (e.g., by the LBS employees or criminals who breached its security) motivates the search for better solutions that would ensure the privacy of user data, and give users control of how their data is being used.

In this thesis, we address this problem using cryptographic techniques. We propose three cryptographic protocols that allow two users to perform proximity testing (check if they are close enough to each other) with the help of two servers.

In the papers 1 and 2, the servers are introduced in order to allow users not to be online at the same time: one user may submit their location to the servers and go offline, the other user coming online later and finishing proximity testing. The drastically improves the practicality of such protocols, since the mobile devices that users usually run may not always be online. We stress that the servers in these protocols merely aid the users in performing the proximity testing, and none of the servers can independently extract the user data.

In the paper 3, we use the servers to offload the users' computation and communication to. The servers here pre-generate correlated random data and send it to users, who can use it to perform a secure proximity testing protocol faster. Paper 3, together with the paper 2, are highly practical: they provide strong security guarantees and are suitable to be executed on resource-constrained mobile devices. In fact, the work of clients in these protocols is close to negligible as most of the work is done by servers.