Differentially-Private Data Aggregation over Encrypted Location Data for Range Counting Query

Location data has the potential to uncover patterns of congestion and overcrowding during specific times of day and days of the week. By pooling location data across different organizations, valuable insights can be derived that would be challenging to obtain independently. For instance, combining binary flag data (0 or 1), such as hotel stays, medical histories, and purchase records, with location data can facilitate range counting to reveal stay trends and the prevalence of infectious diseases in each region. However, the practice of aggregating data from various organizations introduces a critical concern: privacy leakage. When organizations share their data for aggregation, there is a risk that sensitive information could be exposed. To address this privacy challenge, it is imperative to aggregate the data of each organization while preserving privacy, and to make it impossible to infer sensitive information. In this research, we introduce an innovative differentially-private data aggregation protocol, facilitating the analysis of range counting across various organizations while maintaining data encryption throughout the process. Our proposed protocol leverages Homomorphic Encryption to secure both flag data and location information, confidentially merging only shared records to generate a unified table. Subsequently, our approach introduces encrypted noise to the resulting table until Differential Privacy guarantees privacy protection, even upon decryption. However, applying differential privacy to encrypted data carries the risk of enabling adversaries to inject manipulated data at their discretion. To counteract the potential mixing of manipulated and encrypted data, we have developed an algorithm within our proposed protocol to validate the content of encrypted data.