Pair of scientists think they have topped homomorphic encryption
Japanese researchers say they have created a way to perform computations with encrypted data faster, more cheaply and more securely than legacy methods of preserving data privacy and security.
The backdrop of this is the assumption that society 5 running on the Internet of things is on its way and will be the most data-intensive and security-dependent iteration of online life.
A pair of scientists from Tokyo University of Science, say they have a computation method that is better than homomorphic encryption. All of the computations are done on a single server, reportedly without a significant computational cost. Their paper was published late last year in IEEE Access.
The setup, which is supposed to allow for decentralized computation of encrypted data while still working the problem on just one server, is an involved process.
There is a trusted third party, a computing server, four players providing secret inputs to the server and one player restoring the computation output.
The neutral third party creates random numbers, referred to as shares, which are fed to the server and the other players in proscribed combinations. The numbers encrypt the data.
Players perform a computation with the random numbers and generate secret inputs that are sent to the server.
The server uses the shares and secret inputs with new values computed by the third party to perform a series of computations. The result of those computations is sent to a final player who reconstructs the computation result.
“In our proposed method, we realize the advantage of homomorphic encryption without the significant computational cost incurred by homomorphic encryption,” says Professor Keiichi Iwamura, “devising a way to securely handle data.” Iwamura led the study.