With the emergence of cloud computing, it is of critical importance to provide security and privacy guarantees for outsourced computation. This pertains both to providing rigorous assurance that the outsourced computations have been performed according to the considered specification (verifiable computation) and to offering oblivious access to outsourced data (ORAM).
Verifiable Computation. We consider scenarios of a client that remotely stores large amounts of data and later outsources the computation of some function on this data to the server. The client wants to efficiently verify — in time independent of the input-size – the correctness of the result despite no longer knowing the inputs of the outsourced computation, to continuously add data to its remote storage, and to not fix the outsourced function in advance. We propose novel cryptographic techniques that solve the above problem for the class of computations of quadratic polynomials over a large number of variables, covering a wide range of arithmetic computations, in particular, many important statistics. In addition, we address privacy-preserving proofs on authenticated data: a party receives data from a trusted source and is requested to prove correctness of computations over the data to third parties in a privacy-preserving manner..
Oblivious RAM. Oblivious RAM (ORAM) promises oblivious access to central memory, thereby avoiding accidental or intentional leakage of sensitive information and the profiling of user activities. We develop cryptographic systems that simultaneously protect the secrecy and integrity of outsourced data against both untrusted servers and malicious clients, guarantees the anonymity and unlinkability of data accesses, and allows the data owner to selectively grant other clients read and write permissions for outsourced data.