Rational Fair Computation with Computational Sequential Equilibrium

In secure multi-party computation, the property of fairness means that corrupted parties should receive their outputs if and only if the honest parties also receive their outputs. It is hardly achieved when the number of malicious parties is more than half number of the participants. Therefore, fairness is often ignored in secure multi-party computation. Parties in traditional secure computation consist of honest parties, semi-honest parties and malicious parties. Honest parties always follow the protocol. Semi-honest parties also follow the protocol. However, they may keep an internal state of all the corrupted parties attempting to use this to learn private information. Malicious parties arbitrarily deviate from the protocol. Rational parties are neither of them but only want to maximize their utilities. Rational secure computation means that rational parties are allowed to participate in the computation, which opens a new avenue to achieve this desirable property. In this paper, we consider a more realistic rational protocol where parties may have asymmetric information such as utilities and types, which make our protocol distinct with previous ones. On account of these distinctions, a stronger equilibrium named computational sequential equilibrium, which consists of computational sequentially rational and consistent, is put forward to guarantee fairness. At the end of this paper, an simulator is constructed to prove the security of the protocol.