Secure aggregation in a publish-subscribe system

  • Authors:
  • Kazuhiro Minami;Adam J. Lee;Marianne Winslett;Nikita Borisov

  • Affiliations:
  • University of Illinois at Urbana-Champaign, Urbana, IL, USA;University of Pittsburgh, Pittsburgh, PA, USA;University of Illinois at Urbana-Champaign, Urbana, IL, USA;University of Illinois at Urbana-Champaign, Urbana, IL, USA

  • Venue:
  • Proceedings of the 7th ACM workshop on Privacy in the electronic society
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

A publish-subscribe system is an information dissemination infrastructure that supports many-to-many communications among publishers and subscribers. In many publish-subscribe systems, in-network aggregation of input data is considered to be an important service that reduces the bandwidth requirements of the system significantly. In this paper, we present a scheme for securing the aggregation of inputs to such a publish-subscribe system. Our scheme, which focuses on the additive aggregate function, sum, preserves the confidentiality and integrity of aggregated data in the presence of untrusted routing nodes. Our scheme allows a group of publishers to publish aggregate data to authorized subscribers without revealing their individual private inputs to either the routing nodes or the subscribers. In addition, our scheme allows subscribers to verify that routing nodes perform the aggregation operation correctly. We use a message authentication code (MAC) scheme based on the discrete logarithm property to allow subscribers to verify the correctness of aggregated data without receiving the digitally-signed raw data used as input to the aggregation. In addition to describing our secure aggregation scheme, we provide formal proofs of its soundness and safety.