Verifiable computation with reduced informational costs and computational costs

Abstract

Outsourcing computation is a fundamental principle of the new cloud computing paradigm. Among its various aspects, the correctness of the computation result remains paramount. This motivates the birth of Verifiable computation, which aims at efficiently checking the result for general-purpose computation. The common goal of recently sprouted Verifiable computation protocols is to reduce the costs associated with verification at both prover and verifier. Unfortunately, the high computation and communication costs of verification still keep general Verifiable computation away from practicality. Besides the computational costs, we observe that another type of verification cost has been generally ignored until now-the informational costs, namely, the information required for the verification. In particular, in the context of the third-party verification, this cost implies the information leakage of sensitive information regarding the computational task and its results. In this paper, we introduce the new verifiable-computation protocol RIVER, which reduces the computational costs of the verifier and of the prover, comparing to the most recent alternative protocols, and (for the first time in the context of Verifiable computation) addresses and decreases informational costs. © 2014 Springer International Publishing Switzerland.