Attribute-based multi-function verifiable computation

Abstract

Verifiable Computation (VC) allows resource-restricted clients to delegate expensive computations to more powerful servers, and then to verify the correctness of the results. A number of VC schemes have been constructed since it was introduced. However, existing VC schemes support verifying and retrieving results either only by the client (privately) or by anybody (publicly). This one-or-all mode of verification does not fill the need in practical applications, for example, in healthcare applications, some users with special attributes can be authorized to verify and retrieve the results and other users without special attributes cannot be authorized such rights.In this work, we introduce and formalize the notion of attribute-based multi-function verifiable computation (AM-VC). It provides fine-grained access control to the computation results. Specifically, it labels the encoded functions and inputs with policies, as well as associates the secret key of each receiver with attributes. In this way, only the receiver whose attributes satisfy both a function’s policy and an input’s policy can verify and retrieve the corresponding computation result. In addition, executing multiple functions on one encoded input and executing one function on multiple encoded inputs are allowed. We also formally model the security properties of an AM-VC scheme and propose a secure construction.