Real-time and private spatio-temporal data aggregation with local differential privacy

Abstract

Technology and usage advances in wireless communication and smart mobile devices with localization capabilities enable a large number of emerging applications of location-based services, e.g. mobile crowdsourcing applications, which are facilitating our daily life. However, collecting and sharing location data to service providers of applications will give rise to mobile users’ concerns on their privacy, especially location privacy. In this paper, we investigate the problem of real-time spatio-temporal data aggregation with privacy preservation in the local setting. In response to this, we propose a systematic solution based on stringent local differential privacy preservation technique to deal with the problem. Firstly, we introduce a novel definition of (ε, δ)-local differential privacy with the ability to capture the temporal correlation in spatio-temporal data and provide differential privacy preservation. Secondly, we develop an efficient framework of generalized randomized response (GRR) based real-time and private spatio-temporal data aggregation with an untrusted server. Finally, we conduct experiments on two real-world datasets to evaluate our framework. The results show that our GRR based framework significantly outperforms that based on PIM and LRM in data utility and demonstrate its superiority to achieve better trade-off of privacy and utility for real-time spatio-temporal data aggregation with stringent privacy preservation.