Efficient Verification of Input Consistency in Server-Assisted Secure Function Evaluation

We investigate a new model for generic secure computation in which there is a semi-honest server that assists computation between two malicious parties. In this model, we consider a setting in which the server assists multiple executions between two xed players. We present practical and efficient schemes which guarantee that consistent inputs are used by the parties across these multiple executions. We also present efficient schemes where such input consistency can be guaranteed even when one of the parties did not participate in the previous execution.