Secure communication and computation in an insecure environment

Abstract

A mechanism which secures the communication and computation between processors in an insecure distributed environment implements efficient 'compilers' for a protocol between processors. The protocol is one that assures some input-output relation when executed by processors which are not all trusted but with secret and authenticated communication links between every two processors. This protocol is transformed by a compiler into a protocol that guarantees essentially the same input-output relations in the presence of (the same type of) insecure processors and insecure communication links. Additionally, a method maintains secret values for a sequence of periods, each secret value being shared by two or more processors for one or several periods, where the processors are connected by a communication network. Another mechanism establishes different cryptographic keys established for each period of communication. Essentially, the effect of exposures is contained to the period in which they occur, or to a minimal number of following periods, and the effect of exposures is contained to the processors exposed. At each period a processor is called nonfaulty if the adversary does not control it. A processor is called secure at a given period if it is non-faulty and also has a secret key, unknown to the adversary.