Using Sparse Matrices to Prevent Information Leakage in Cloud Computing

Abstract

Cloud computing represents the promise of outsourcing of scientific computing such as matrix multiplication. However, this can introduce new vulnerabilities such as information leakage. Cloud server intentionally or unintentionally may reveal sensitive input matrices of the client as well as multiplication results to unauthorised entities. In this paper, we propose two protocols that use sparse matrices to prevent information leakage in outsourcing matrix multiplications to cloud computing without encryption. The protocols are considered lightweight compared to other comparable approaches. We also provide a running example to demonstrate how the protocols ensure no information leakage of client data.