A Data Ownership Privacy Provider Framework in Cloud Computing

Abstract

The complementary problem of secure storage of private cloud data has been studied extensively in the literature but cannot usually be applied while the data is in decrypted form for the duration of a computation. Secure multiparty computation and differential privacy are both powerful approaches to privacy preserving cloud computation on decrypted data, but are inapplicable to many real world cloud computations. In particular, jobs submitted to the cloud as arbitrary binary code are difficult to automatically reformulate as secure multiparty computations, and high differential privacy sometimes comes at the expense of highly imprecise, noisy results. In these cases, the level of privacy can sometimes be improved by concealing data ownership, provenance, and/or semantics from the participants in a computation in addition to or instead of anonymizing the data itself.