Privacy-preserving Data Aggregation for Big Data in Financial Institutions

Data aggregation plays an important role in the era of big data. Financial institutions aggregate user data to obtain the global information about their users. At the same time, they intend to hide the detailed data from third parties such as fusion center when participating in the aggregation due to concerns on data privacy. Based on the secure multi-party computation and pseudonym, in this paper, we propose a privacy preserving data aggregation mechanism. In this new mechanism, users are grouped into pairs anonymously, and their data is divided and mixed between pairwise users. Compared with the existing methods, this new scheme can achieve lower losses when link failure occurs, and resist the collusion attacks from users.