Multiparty computations in varying contexts

Abstract

Recent developments in the automatic transformation of protocols into Secure Multiparty Computation (SMC) interactions, and the selection of appropriate schemes for their implementation have improved usabililty of SMC. Poor performance along with data leakage or errors caused by coding mistakes and complexity had hindered SMC usability. Previous practice involved integrating the SMC code into the application being designed, and this tight integration meant the code was not reusable without modification. The progress that has been made to date towards the selection of different schemes focuses solely on the two-party paradigm in a static set-up, and does not consider changing contexts. Contexts, for secure multiparty computation, include the number of participants, link latency, trust and security requirements such as broadcast, dishonest majority etc. Variable Interpretation is a concept we propose whereby specific domain constructs, such as multiparty computation descriptions, are explicitly removed from the application code and expressed in SMC domain representation. This mirrors current practice in presenting a language or API to hide SMC complexity, but extends it by allowing the interpretation of the SMC to be adapted to the context. It also decouples SMC from human co-ordination by introducing a rule-based dynamic negotiation of protocols. Experiments were carried out to validate the method, running a multiparty computation on a variable interpreter for SMC using different protocols in different contexts.