Assisting server for secure multi-party computation

  • Authors:
  • Jens-Matthias Bohli;Wenting Li;Jan Seedorf

  • Affiliations:
  • NEC Laboratories Europe, Heidelberg, Germany;NEC Laboratories Europe, Heidelberg, Germany;NEC Laboratories Europe, Heidelberg, Germany

  • Venue:
  • WISTP'12 Proceedings of the 6th IFIP WG 11.2 international conference on Information Security Theory and Practice: security, privacy and trust in computing systems and ambient intelligent ecosystems
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Distributed threats like botnets are among the most serious threats in the Internet. Due to their distributed nature, these attacks are difficult to detect in an early stage without the collaboration of several network operators. However, the exchange of monitoring data between different parties turns out to be difficult in practice, due to the desire of operators not to disclose network internals and legal data protection requirements. Secure Multi-Party Computation (SMC) for privacy-preserving sharing of network monitoring data can be a solution to the problem. As real-time performance of SMC is important for this application, we investigate ways to speed up SMC. The focus and contribution of our work is a new model for SMC that enables to increase the performance of certain SMC primitives significantly. We introduce an assisting server which operates on dedicated, intermediate data values in plaintext. The overall rationale behind our approach is that the performance gains outweigh the slight decrease in security introduced by revealing intermediate computation results to the assisting server. We propose a new primitive for checking the equality between two values, equal+, based on our new model. Through prototypical implementation we compare equal+ with existing algorithms. Further, we evaluate equal+ in the context of a cooperative network monitoring application, link-counting. Our results demonstrate that certain SMC applications can be computed much faster with our approach. Finally, we discuss the security implications of the new model.