InnerCircle: A parallelizable decentralized privacy-preserving location proximity protocol

Abstract

Abstract-Location Based Services (LBS) are becoming increasingly popular. Users enjoy a wide range of services from tracking a lost phone to querying for nearby restaurants or nearby tweets. However, many users are concerned about sharing their location. A major challenge is achieving the privacy of LBS without hampering the utility. This paper focuses on the problem of location proximity, where principals are willing to reveal whether they are within a certain distance from each other. Yet the principals are privacy-sensitive, not willing to reveal any further information about their locations, nor the distance. We propose InnerCircle, a novel secure multi-party computation protocol for location privacy, based on partially homomorphic encryption. The protocol achieves precise fully privacy-preserving location proximity without a trusted third party in a single round trip. We prove that the protocol is secure in the semihonest adversary model of Secure Multi-party Computation, and thus guarantees the desired privacy properties. We present the results of practical experiments of three instances of the protocol using different encryption schemes. We show that, thanks to its parallelizability, the protocol scales well to practical applications.

Faster Privacy-Preserving Location Proximity Schemes

“…In this section, we compare our proposed protocols with the state-of-the-art PPLP protocol of Hallgren et al [15,16]. We instantiate all primitives in our PPLP protocols to achieve a security level of κ = 128 bits.…”

“…So far several solutions for privacy-preserving location proximity (PPLP) schemes have been proposed, e.g., [6,16,25,[29][30][31][35][36][37]. In early literature [6], privacy-preserving location proximity computation is realized by an imprecise location-based range query that allows a user to approximately learn if any of its communication partners is within a fixed distance from her current location.…”

“…Hallgren et al [16] develop a PPLP protocol built on only AHPKE to test whether two users' locations are within a given distance threshold (without a trusted third party). Their construction makes use of a similar distance obfuscation method as in [36].…”

PILOT: Practical Privacy-Preserving Indoor Localization Using OuTsourcing

“…distances ← d ABOY m (RSS C , RefPoints, N) 1 : distances ← ∅ 2 : foreach RSSS in RefPoints do 3 : dist A ← 0 4 : for i = 1 : N do 5 : : while (j < coords.size()) do 10 : if ( coords[j] Y == NULL) 11 : j ← j + 2 · i 12 : continue 13 : else if ( coords[j + i] Y == NULL) 14 : j ← j + 2 · i 15 : continue 16 : else 17 :…”

“…Privacy is a major concern in LBSs because the location history allows very accurate user profiling and even predicting users' future movements [4]. A lot of literature exists on privacy-preserving LBSs, e.g., [5,6,7,8,9]. Typically in these works, the user's location is assumed to be known by the user and the privacy concerns are about how this location information is shared and used in an LBS.…”