Federated Analytics: Opportunities and Challenges

In this article, we present federated analytics, a new distributed computing paradigm for data analytics applications with privacy concerns. With the advances of sensing, communication, and edge computing technologies, data are massively generated, transmitted and analyzed in an edge-cloud computing environment. In many applications, the edge devices and the data generated in the edge belong to heterogeneous owners. Data privacy and confidentiality have become increasing concerns to these owners. The current edge-cloud computing paradigm for data analytics, where data are sent to a central server for analytics, can no longer match the application requirements. Federated analytics is a newly proposed computing paradigm where raw data are kept local with local analytics and only the insights generated from local analytics are sent to a server for result aggregation. Federated analytics differs from the recent federated learning paradigm in the sense that federated learning emphasizes collaborative model training, whereas federated analytics emphasizes drawing conclusions from data. In this article, we first clarify what federated analytics is and its position in the research literature. We then present why we need federated analytics, that is, the motivation and application case studies. Finally, we discuss the opportunities and challenges of federated analytics.

Automated summary

Federated analytics is a new distributed computing paradigm focusing on local data privacy and insight aggregation. It differs from federated learning by emphasizing drawing conclusions from data. The article discusses the definition, motivation, application case studies, as well as the opportunities and challenges of federated analytics.