Performance Impact Analysis of Rounds and Amounts of Communication in Secure Multiparty Computation Based on Secret Sharing

Abstract

Data security is a key element in many IT systems. Secret values can be stored in databases after encrypting them (in particular by applying a secret sharing schema) but, in practice, these confidential values often need to be processed otherwise losing their usefulness. A possible solution is secure multiparty computation (known as MPC) that allows the computation of various functions directly on encrypted data, thus guaranteeing the confidentiality of the information during the whole process of querying databases. The main drawback for such systems is performance degradation, as secure computations on secret data imply multiple communication rounds and the exchange of messages with different lengths. In this paper, two sets of MPC subprotocols are benchmarked and compared considering the number of rounds and the communication cost, in order to assess which one of these parameters has a bigger impact on performance. Our results show that a small cost is more important in terms of efficiency than reducing the number of rounds.