Secure computation research has gained traction internationally in the last five years. In the United States, the DARPA PROCEED program (2011-2015) focused on development of multiple SC paradigms and improving their performance. In the European Union, the PRACTICE program (2013-2016) focuses on its use to secure cloud computing. Both programs have demonstrated exceptional prototypes and performance improvements. In this paper, we collect the results from both programs and other published literature to present the state of the art in what can be achieved with today's secure computing technology. We consider linear secret sharing based computations, garbled circuits and fully homomorphic encryption. We describe theoretical and practical criteria that can be used to characterize secure computation paradigms and provide an overview of common benchmarks such as AES evaluation.
↧