In this project, the PIs study secure computation outsourcing in cloud computing with the focus on widely applicable engineering computing and optimization problems. Their methodology is to explicitly decompose computations into public programs and private data and leverage the structures of specific computations for achieving desirable trade-offs among security, efficiency, and practicality.
The PIs propose to organize the mechanisms into a hierarchy where computation can be represented at various abstraction levels, and then explore a systematic methodology consisting of the following three methods: (1) problem transformations that encrypt the data such that the computation can be performed on the same abstraction level, (2) procedure transformations that leverage the mechanisms defined at a lower abstraction level as a subroutine for secure computation outsourcing, and (3) structural-preserving transformations that further improve the practical efficiency of mechanisms by maintaining favorable problem structures.
The PIs expect the outcomes of this research to be adopted by application developers, who will build applications to support secure computation outsourcing either privately for end-users within the same organization, or for public end-users resembling the practices of software-as-a-service (SaaS).
