Distributed privacy-preserving data aggregation via anonymization

Data aggregation is a key element in many applications that draw insights from data analytics, such as medical research, smart metering, recommendation systems and real-time marketing. In general, data is gathered from several sources, processed, and publicly released for data analysis. Since the considered data might contain personal and sensitive information, special handling of private data is required. In this paper, we present a novel distributed privacy-preserving data aggregation protocol, called ADiPA. It relies on anonymization techniques for protecting personal data, such as k-anonymity, l-diversity and t-closeness. Its purpose is to allow a set of entities to derive aggregate results from data tables that are partitioned across these entities in a fully decentralized manner while preserving the privacy of their individual sensitive inputs. ADiPA neither relies on a trusted third party nor on cryptographic techniques. The protocol performs accurate aggregation when communication links and nodes do not fail. © Springer International Publishing Switzerland 2015.