Secure multiparty computation


Jump to: navigation, search

This page is under construction. Do not rely on its accuracy until it is finished. Please edit this page or use the talk page to leave comments.

Secure Multiparty Computation (MPC) enables mutiple parties to compute a function f without revealing information about the inputs to each other. For most discussion, we consider the two party case with Alice and Bob, each with inputs to the function f, x and y.

Some examples where MPC is useful:

  1. Medical record datamining - Hospitals wish to perform data mining on their records and combine with other hospitals, although they can not share any patient information due to privacy policy.
  2. A Card game with no dealer - Alice and Bob want to play an honest card game with no dealer, although neither Alice or Bob trust that the other will not cheat.
  3. An online Auction system without a trusted third party.

We define the security of MPC by setting the ideal goal to be performing computation as if there was a trusted third party. In the two party case, Alice and Bob simply exchange their information with the third party who then performs the computation and returns the result. In the real world there is no trusted third party, instead Alice and Bob communicate directly with each other (in the two party case).

Left Right

When dealing with MPC we will make some restrictions on the Adversary in Standalone security and discuss an algorithm called Yao's garbled circuit for the two party restricted adversary case. Other algorithms for more general adversaries do exist.

Personal tools