With computation power in the cloud becoming a commodity, it is more and more convenient to outsource computations to external computation parties. Assuring confidentiality, even of inputs by mutually distrusting inputters, is possible by distributing computations between different parties using multiparty computation. Unfortunately, this typically only guarantees correctness if a limited number of computation parties are malicious. If correctness is needed when all computation parties are malicious, then one currently needs either fully homomorphic encryption or ``universally verifiable'' multiparty computation; both are impractical for large computations. In this paper, we show for the first time how to achieve practical privacy-friendly outsourcing with correctness guarantees, by using normal multiparty techniques to compute the result of a computation, and then using slower verifiable techniques only to verify that this result was correct. We demonstrate the feasibility of our approach in a linear programming case study.
Keywords: secret sharing , threshold cryptography, zero knowledge
|Number of pages
|Published - 2015
|Cryptology ePrint Archive