Privc: Privacy Preserving Verifiable Computation

Abstract

Verifiable computation in cloud setup with a privacy-preserving feature is an interesting research problem which can find application in monitoring system such as healthcare application. For example, a public cloud service provider can facilitate computation service on patients data or symptoms based on doctors prescription (e.g., prediction function) without letting the cloud server know on anything about patient's data and the patient can verify the computed response. We present a privacy-preserving verifiable computation (PriVC) scheme in which the server can authenticate a user in a privacy-preserving manner, and compute on encrypted data of the user stored in the public cloud. The server provides the proof of computation which can be verified by the user. The PriVC preserves the privacy of the user and ensures undeniability of the service offered as well as the service consumed. The PriVC scheme uses homomorphic encryption for user data encryption and a private polynomial function for computation on encrypted data. We show that the PriVC scheme is secure under indistinguishability against chosen function attack (IND-CFA), and the proof of computation is unforgeable in the standard model.