Techniques to protect privacy against inference attacks in location based services

In this paper, we study potential inference attacks targeting Location Based Service (LBS) users, and provide heuristic defense techniques to protect their privacy against such attacks. Having access to supplemental information such as subsequent query times, speed limits/travel times on the underlying road-network, and/or the residential/commercial address directory, adversaries might be able to infer sensitive information such as location, identity, and/or lifestyle about the querying LBS user. To prevent adversaries from connecting external information to user queries, we apply various heuristic privacy-preserving algorithms whose objective is to alter user queries in order to protect users against inference attacks while providing exact results in a timely manner. Our algorithms enable users to customize their privacy levels based on individual's preferences through the use of flexible user-controlled parameters. For this, we introduce the novel notion of (i, j)-privacy. We evaluate our algorithms experimentally on different road-networks varying a number of input parameters and present the results here. The outcomes of our experiments confirm that except for special cases where a high anonymity level is requested or queries are submitted with very high frequency, our algorithms provide quality results in less than few seconds