UaESMC: Usable and Efficient Secure Multiparty Computation FET OPEN 3rd Year Project revision

Secure Multiparty Computation (SMC) is a universal cryptographic functionality. In this functionality, there are n parties P1, . . . , Pn, each holding a piece of data xi, and wishing to compute a function (y1, . . . , yn) = f(x1,...,xn). They require that each party Pi only learns yi (and anything deducible from the fact that on Pi’s input xi, the output was yi), but nothing beyond that. The universality of SMC stems from the fact that any e↵ciently computable function can serve in place of f. In the 1980s, it was shown how to transform the description of any such f to a cryptographic protocol that implements SMC for that f. The objective of UaESMC is to bring the benefits of SMC to many more and di↵erent fields of activity, where currently the parties have to weigh the benefits obtained from the interaction with other parties against the privacy losses this interaction entails. We are looking for sets of techniques and procedures that convince the parties to adopt SMC and interact without the need to consider the sensitivity of the data they would have to input to a joint computation. Regarding the techniques, we need reasonably ecient SMC protocols for the computational tasks that are actually relevant in practice. Regarding the procedures, we need means to achieve accountability and truthfulness despite strong privacy guarantees. It makes sense to perform a computation only if its result is useful, which requires that the inputs submitted by di↵erent parties are actually their true ones. This requires the design of appropriate incentive mechanisms.