Distributed, Privacy-Aware Location Data Aggregation

Analyzing location data from many individuals can provide valuable insights, especially when linked with private attributes like personal health information. A recent application includes identifying COVID-19 outbreaks by aggregating individuals’ health data across a geographical hierarchy. However, analyzing such sensitive information can threaten the individuals’ privacy, especially when honest-but-curious third parties are involved. To encourage people to share their data for such analyses, strong privacy protection and building trust in the privacy approach are crucial, requiring clear privacy parameters that can be tailored to individual needs. To address these requirements, we introduce DIPALDA, a new anonymization technique for DIstributed, Privacy-Aware Location Data Aggregation on hierarchically structured personal location data. DIPALDA leverages three privacy parameters: k-anonymity, minimum cloaking area size, and maximum re-identification probability, effectively countering re-identification and location privacy attacks. Our extensive experiments with COVID-19 propagation data demonstrate that DIPALDA achieves a suitable trade-off between utility, privacy, and explainability.