Verifiable local computation on distributed data
Liang Feng Zhang, Reihaneh Safavi–Naini, Xiao Wei Liu · 2014
Outsourcing of storage and computation from a resource-restricted client to a powerful cloud raises many security issues such as the privacy of the data and the integrity of any delegated computation on the outsourced data. While many techniques have been introduced to protect the client's data privacy or computation integrity, achieving both of them is challenging. In this paper, we propose a multi-server verifiable local computation (VLC) model where the client can privately outsource data blocks m=(m1, ..., mn) to cloud servers and later verify computations on any portion of the outsourced data. We propose two constructions of multi-server VLC schemes. Our schemes achieve data privacy in the sense that no collusion of a subset (size less than a threshold) of the cloud servers can learn any information about $m$; and computation integrity in the sense that no collusion of a subset (size less than a threshold) of the cloud servers can cause the client to accept their incorrect answers. Our first construction is solely based on PRF and very efficient; our second construction uses bilinear maps and achieves amortized closed-form efficiency over multiple computations of a function.