TACO: A Tree-based Approach to Customizing Location Obfuscation based on User Policies

A large body of literature exists for studying Location obfuscation in different contexts. However, the obfuscation functions generated by existing systems are not easily customizable by end users. Users might find it difficult to understand the parameters involved (e.g., obfuscation range and granularity of location representation) and set realistic trade-offs between privacy and utility. In this paper, we propose a new framework called, TACO, i.e., Tree-based Approach to Customizing location Obfuscation, which can generate location obfuscation functions that provide strong privacy guarantees while being easily customizable via user-specified policies. First, we develop a semantic representation of a given region using tree structure. These data structures assist users in specifying their privacy requirements using policies. Second, we design a rigorous privacy model based on Geo-Indistinguishability for TACO using this tree structure. Third, we implement enforcement techniques in TACO to translate user policies to appropriate parameters and generate a robust, customized obfuscation function for each user. Finally, we carry out experiments on real world datasets to evaluate the effectiveness of the framework under different settings.