Edge Computing Enhanced Privacy Preserving for Location Based Services
Lichuan Ma, Qingqi Pei, Huizi Xiao, Hongning Li, Zi Li, Kefeng Fan · 2019
Recently, location based services have brought much convenience in our daily life and thus gained much attentions from academic and industrial areas. However, some sensitive information, like the locations and query contents can be utilized to analyze the behavior pattern, preferences, habits and interests of the users. The leakage of such sensitive information might lead to the occurrence of spam advertising or even personal injuries. As a result, preserving privacy for the users is of great significance to motivate LBSs to go further. Concurrently, the privacy preserving methods for LBSs can be classified into three categories: anonymity-based, obfuscation-based, dummybased, and homomorphic encryption-based. Among them, the anonymity-based approaches are fragile to tracking attacks. The obfuscation-based and dummy-based approaches sacrifice service accuracy for the privacy-preserving goal. The homomorphic encryption-based approaches can provide stronger security guarantee without degrading the service accuracy. But this kind of approaches would cause heavy computation and communication overheads. Fortunately, by offloading complex ciphertext computations to the edge nodes, the edge computing paradigm offers a new thought to practically implement homomorphic encryption-based approaches for privacy preserving in LBSs. Thus in this paper, we propose an edge computing enhanced privacy preserving framework for location based services. The edge node is responsible for computing the virtual location of any user who inspires a LBS query via K-anonymity technology. On the basis of ciphertext computations, the service response is constructed via the interactions between the edge node and the LBS server. The basic idea is that the encrypted locations are stored in the edge node and the LBS server can only get the rank of the distances between the virtual location and service targets corresponding to the query content. Security analysis and extensive experiments verify the effectiveness of the proposed method.